diff --git a/.gitignore b/.gitignore index 84df49b9c..c9e20119f 100644 --- a/.gitignore +++ b/.gitignore @@ -157,3 +157,5 @@ auto3dseg/notebooks/datalist.json *.png *.np* *.pt +competitions/kaggle/Cryo-ET/1st_place_solution/data/ +competitions/kaggle/Cryo-ET/1st_place_solution/results/ diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/README.md b/competitions/kaggle/Cryo-ET/1st_place_solution/README.md new file mode 100644 index 000000000..4ea71eaeb --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/README.md @@ -0,0 +1,117 @@ +## Introduction + +This tutorial illustrates how to use MONAI for cryo electron tomography. The pipeline and models were partly used to win the [Cryo-ET competition on kaggle](https://www.kaggle.com/competitions/czii-cryo-et-object-identification/overview). The tutorial was tested with nvidia/pytorch:24.08-py3 docker container and a single A100 GPU. + +## What is Cryo-ET? + +If you ask ChatGPT: + +Cryo-ET (Cryo-Electron Tomography) is an advanced imaging technique that allows scientists to visualize biological structures in near-native states at high resolution. It combines cryogenic sample preservation with electron tomography to generate three-dimensional (3D) reconstructions of cellular structures, protein complexes, and organelles. + +### How It Works +1. Cryo-Fixation: The sample (e.g., a cell or a purified macromolecular complex) is rapidly frozen using liquid ethane or similar methods to prevent ice crystal formation, preserving its natural state. +2. Electron Microscopy: The frozen sample is placed under a transmission electron microscope (TEM), where images are taken from multiple angles by tilting the sample. +3. Tomographic Reconstruction: Computational algorithms combine these 2D images to create a detailed 3D model of the structure. + +### Applications +Studying cellular architecture at nanometer resolution. +Visualizing macromolecular complexes in their native environments. +Understanding interactions between viruses and host cells. +Investigating neurodegenerative diseases, cancer, and infectious diseases. +Cryo-ET is particularly powerful because it enables direct imaging of biological systems without the need for staining or chemical fixation, preserving their native conformation. + +## Requirements + +- docker +- git +- kaggle API credentials + +# Running the tutorial + +1. Download the tutorial code from the ProjectMONAI repository. + +```bash +git clone https://github.com/Project-MONAI/tutorials.git +cd tutorials/competitions/kaggle/Cryo-ET/1st_place_solution/ +``` + +2. Run container to start the tutorial. + +```bash +docker run --gpus all --ipc=host --ulimit memlock=-1 --ulimit stack=67108864 -v $PWD:/workspace/ -it nvcr.io/nvidia/pytorch:24.08-py3 /bin/bash +``` + +if you want to use the kaggle API to download the data, you need to mount your kaggle.json file into the container. You can do this by adding the following flag to the docker run command: + +```bash +docker run --gpus all --ipc=host --ulimit memlock=-1 --ulimit stack=67108864 -v $PWD:/workspace/ -v $HOME/.config/kaggle/:/root/.kaggle -it nvcr.io/nvidia/pytorch:24.08-py3 /bin/bash +``` + +3. Install necessary additional pip packages inside the container by running the following command on the prompt you get after running the previous command: + + +```bash +pip install -r requirements.txt +``` + +4. Download the data + +This tutorial is build upon the official Cryo ET competition data. +It can be downloaded to a local ```DATA_FOLDER``` directly from kaggle (You will also need to follow the competition url and click "join competition" to accept the terms and conditions): https://www.kaggle.com/competitions/czii-cryo-et-object-identification/data . + +Alternativly it can be downloaded using the kaggle API (which can be installed via ```pip install kaggle```). If you decide to use the Kaggle API you need to create a Kaggle account and configure your token as described [here](https://github.com/Kaggle/kaggle-api/blob/main/docs/README.md#api-credentials) and then be allowed to download the data with the following command: + +```bash +export DATA_FOLDER=$PWD/data +mkdir -p $DATA_FOLDER +kaggle competitions download -c czii-cryo-et-object-identification -p $DATA_FOLDER +``` + +Unzip the competition dataset to DATA_FOLDER + +```bash +cd $DATA_FOLDER +unzip czii-cryo-et-object-identification.zip -d czii-cryo-et-object-identification/ +``` + +If you change the DATA_FOLDER location, have to adjust path to the `cfg.data_folder` data at `configs/common_config.py`. + +## Training models + +For the competition we created a cross-validation scheme by simply simply splitting the 7 training tomographs into 7 folds. I.e. we train on 6 tomographs and use the 7th as validation. +For convenience we provide a file ```train_folded_v1.csv``` which contains the original training annotations and was also extended by a column containing fold_ids. + +We solve the competition with a 3D-segmentation approach leveraging [MONAI's FlexibleUNet](https://docs.monai.io/en/stable/networks.html#flexibleunet) architecture. Compared to the original implementation we adjusted the network to output more featuremap and enable deep-supervision. The following illustrates the resulting architecture at a high level: + +

+ figure of a Partly UNet + +We provide three different configurations which differ only in the used backbone and output feature maps. The configuration files are .py files and located under ```configs``` and share all other hyper-parameters. Each hyperparameter can be overwriten by adding a flag to the training command. To train a resnet34 version of our segmentation model simply run + +```bash +export RESULTS=$PWD/results +mkdir -p $RESULTS +python train.py -C cfg_resnet34 --output_dir $RESULTS +``` + +This will save checkpoints under the specified $RESULTS when training is finished. +By default models are trained using bfloat16 which requires a GPU capable of that. Alternatively you can set ```cfg.bf16=False``` or overwrite as flag ```--bf16 False``` when running ```train.py ```. + +### Replicating 1st place solution (segmentation part) + +To train checkpoints necessary for replicating the segmentation part of the 1st place solution run training of 2x fullfits for each model. Thereby ```cfg.fold = -1``` results in training on all data, and using ```fold 0``` as validation. +``` +python train.py -C cfg_resnet34 --fold -1 +python train.py -C cfg_resnet34 --fold -1 +python train.py -C cfg_resnet34_ds --fold -1 +python train.py -C cfg_resnet34_ds --fold -1 +python train.py -C cfg_effnetb3 --fold -1 +python train.py -C cfg_effnetb3 --fold -1 +``` + +## Inference + +Inference after models are converted with torch jit is shown in our 1st place submission kaggle kernel. + +https://www.kaggle.com/code/christofhenkel/cryo-et-1st-place-solution?scriptVersionId=223259615 diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/configs/cfg_effnetb3.py b/competitions/kaggle/Cryo-ET/1st_place_solution/configs/cfg_effnetb3.py new file mode 100644 index 000000000..82d7ef8e0 --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/configs/cfg_effnetb3.py @@ -0,0 +1,34 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +from common_config import basic_cfg +import os +import pandas as pd +import numpy as np +import monai.transforms as mt + +cfg = basic_cfg + +cfg.name = os.path.basename(__file__).split(".")[0] +cfg.output_dir = f"/mount/cryo/models/{os.path.basename(__file__).split('.')[0]}" + +# model +cfg.backbone = "efficientnet-b3" +cfg.backbone_args = dict( + spatial_dims=3, + in_channels=cfg.in_channels, + out_channels=cfg.n_classes, + backbone=cfg.backbone, + pretrained=cfg.pretrained, +) +cfg.class_weights = np.array([64, 64, 64, 64, 64, 64, 1]) +cfg.lvl_weights = np.array([0, 0, 0, 1]) diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/configs/cfg_resnet34.py b/competitions/kaggle/Cryo-ET/1st_place_solution/configs/cfg_resnet34.py new file mode 100644 index 000000000..470d69087 --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/configs/cfg_resnet34.py @@ -0,0 +1,35 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from common_config import basic_cfg +import os +import pandas as pd +import numpy as np + +cfg = basic_cfg + +# paths +cfg.name = os.path.basename(__file__).split(".")[0] +cfg.output_dir = f"/mount/cryo/models/{os.path.basename(__file__).split('.')[0]}" + + +# model + +cfg.backbone = "resnet34" +cfg.backbone_args = dict( + spatial_dims=3, + in_channels=cfg.in_channels, + out_channels=cfg.n_classes, + backbone=cfg.backbone, + pretrained=cfg.pretrained, +) +cfg.class_weights = np.array([256, 256, 256, 256, 256, 256, 1]) +cfg.lvl_weights = np.array([0, 0, 0, 1]) diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/configs/cfg_resnet34_ds.py b/competitions/kaggle/Cryo-ET/1st_place_solution/configs/cfg_resnet34_ds.py new file mode 100644 index 000000000..6b177713a --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/configs/cfg_resnet34_ds.py @@ -0,0 +1,32 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from common_config import basic_cfg +import os +import pandas as pd +import numpy as np + +cfg = basic_cfg + +# paths +cfg.name = os.path.basename(__file__).split(".")[0] +cfg.output_dir = f"/mount/cryo/models/{os.path.basename(__file__).split('.')[0]}" + +cfg.backbone = "resnet34" +cfg.backbone_args = dict( + spatial_dims=3, + in_channels=cfg.in_channels, + out_channels=cfg.n_classes, + backbone=cfg.backbone, + pretrained=cfg.pretrained, +) +cfg.class_weights = np.array([64, 64, 64, 64, 64, 64, 1]) +cfg.lvl_weights = np.array([0, 0, 1, 1]) diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/configs/common_config.py b/competitions/kaggle/Cryo-ET/1st_place_solution/configs/common_config.py new file mode 100644 index 000000000..efcddd574 --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/configs/common_config.py @@ -0,0 +1,209 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +from types import SimpleNamespace +from monai import transforms as mt + +cfg = SimpleNamespace(**{}) + +# stages +cfg.train = True +cfg.val = True +cfg.test = True +cfg.train_val = True + +# dataset +cfg.batch_size_val = None +cfg.use_custom_batch_sampler = False +cfg.val_df = None +cfg.test_df = None +cfg.val_data_folder = None +cfg.train_aug = None +cfg.val_aug = None +cfg.data_sample = -1 + +# model + +cfg.pretrained = False +cfg.pretrained_weights = None +cfg.pretrained_weights_strict = True +cfg.pop_weights = None +cfg.compile_model = False + +# training routine +cfg.fold = 0 +cfg.optimizer = "Adam" +cfg.sgd_momentum = 0 +cfg.sgd_nesterov = False +cfg.lr = 1e-4 +cfg.schedule = "cosine" +cfg.num_cycles = 0.5 +cfg.weight_decay = 0 +cfg.epochs = 10 +cfg.seed = -1 +cfg.resume_training = False +cfg.distributed = False +cfg.clip_grad = 0 +cfg.save_val_data = True +cfg.gradient_checkpointing = False +cfg.apex_ddp = False +cfg.synchronize_step = True + +# eval +cfg.eval_ddp = True +cfg.calc_metric = True +cfg.calc_metric_epochs = 1 +cfg.eval_steps = 0 +cfg.eval_epochs = 1 +cfg.save_pp_csv = True + + +# ressources +cfg.find_unused_parameters = False +cfg.grad_accumulation = 1 +cfg.syncbn = False +cfg.gpu = 0 +cfg.dp = False +cfg.num_workers = 8 +cfg.drop_last = True +cfg.save_checkpoint = True +cfg.save_only_last_ckpt = False +cfg.save_weights_only = False + +# logging, +cfg.neptune_project = None +cfg.neptune_connection_mode = "debug" +cfg.save_first_batch = False +cfg.save_first_batch_preds = False +cfg.clip_mode = "norm" +cfg.data_sample = -1 +cfg.track_grad_norm = True +cfg.grad_norm_type = 2.0 +cfg.track_weight_norm = True +cfg.norm_eps = 1e-4 +cfg.disable_tqdm = False + + +# paths + +cfg.data_folder = "/workspace/data/czii-cryo-et-object-identification/train/static/ExperimentRuns/" +cfg.train_df = "train_folded_v1.csv" + + +# stages +cfg.test = False +cfg.train = True +cfg.train_val = False + +# logging +cfg.neptune_project = None +cfg.neptune_connection_mode = "async" + +# model +cfg.model = "mdl_1" +cfg.mixup_p = 1.0 +cfg.mixup_beta = 1.0 +cfg.in_channels = 1 +cfg.pretrained = False + +# data +cfg.dataset = "ds_1" +cfg.classes = ["apo-ferritin", "beta-amylase", "beta-galactosidase", "ribosome", "thyroglobulin", "virus-like-particle"] +cfg.n_classes = len(cfg.classes) + +cfg.post_process_pipeline = "pp_1" +cfg.metric = "metric_1" + + +cfg.particle_radi = { + "apo-ferritin": 60, + "beta-amylase": 65, + "beta-galactosidase": 90, + "ribosome": 150, + "thyroglobulin": 130, + "virus-like-particle": 135, +} + +cfg.voxel_spacing = 10.0 + + +# OPTIMIZATION & SCHEDULE + +cfg.fold = 0 +cfg.epochs = 10 + +cfg.lr = 1e-3 +cfg.optimizer = "Adam" +cfg.weight_decay = 0.0 +cfg.warmup = 0.0 +cfg.batch_size = 8 +cfg.batch_size_val = 16 +cfg.sub_batch_size = 4 +cfg.roi_size = [96, 96, 96] +cfg.train_sub_epochs = 1112 +cfg.val_sub_epochs = 1 +cfg.mixed_precision = False +cfg.bf16 = True +cfg.force_fp16 = True +cfg.pin_memory = False +cfg.grad_accumulation = 1.0 +cfg.num_workers = 8 + + +# Saving +cfg.save_weights_only = True +cfg.save_only_last_ckpt = False +cfg.save_val_data = False +cfg.save_checkpoint = True +cfg.save_pp_csv = False + + +cfg.static_transforms = static_transforms = mt.Compose( + [ + mt.EnsureChannelFirstd(keys=["image"], channel_dim="no_channel"), + mt.NormalizeIntensityd(keys="image"), + ] +) +cfg.train_aug = mt.Compose( + [ + mt.RandSpatialCropSamplesd(keys=["image", "label"], roi_size=cfg.roi_size, num_samples=cfg.sub_batch_size), + mt.RandFlipd( + keys=["image", "label"], + prob=0.5, + spatial_axis=0, + ), + mt.RandFlipd( + keys=["image", "label"], + prob=0.5, + spatial_axis=1, + ), + mt.RandFlipd( + keys=["image", "label"], + prob=0.5, + spatial_axis=2, + ), + mt.RandRotate90d( + keys=["image", "label"], + prob=0.75, + max_k=3, + spatial_axes=(0, 1), + ), + mt.RandRotated( + keys=["image", "label"], prob=0.5, range_x=0.78, range_y=0.0, range_z=0.0, padding_mode="reflection" + ), + ] +) + +cfg.val_aug = mt.Compose([mt.GridPatchd(keys=["image", "label"], patch_size=cfg.roi_size, pad_mode="reflect")]) + + +basic_cfg = cfg diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/data/ds_1.py b/competitions/kaggle/Cryo-ET/1st_place_solution/data/ds_1.py new file mode 100644 index 000000000..1c56f7563 --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/data/ds_1.py @@ -0,0 +1,102 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +import os +import torch +import numpy as np +from torch.utils.data import Dataset, DataLoader +import zarr +from tqdm import tqdm + + +def batch_to_device(batch, device): + batch_dict = {key: batch[key].to(device) for key in batch} + return batch_dict + + +def collate_fn(batch): + + keys = batch[0].keys() + batch_dict = {key: torch.cat([b[key] for b in batch]) for key in keys} + return batch_dict + + +tr_collate_fn = collate_fn +val_collate_fn = collate_fn + +import monai.data as md +import monai.transforms as mt + + +class CustomDataset(Dataset): + def __init__(self, df, cfg, aug, mode="train"): + + self.cfg = cfg + self.mode = mode + self.df = df + self.experiment_df = self.df.drop_duplicates(subset="experiment").copy() + self.exp_dict = self.df.groupby("experiment") + self.class2id = {c: i for i, c in enumerate(cfg.classes)} + self.n_classes = len(cfg.classes) + self.data_folder = cfg.data_folder + + self.random_transforms = aug + data = [self.load_one(img_id) for img_id in tqdm(self.experiment_df["experiment"].values)] + data = md.CacheDataset(data=data, transform=cfg.static_transforms, cache_rate=1.0) + + if self.mode == "train": + self.monai_ds = md.Dataset(data=data, transform=self.random_transforms) + self.sub_epochs = cfg.train_sub_epochs + self.len = len(self.monai_ds) * self.sub_epochs + else: + self.monai_ds = md.CacheDataset(data=data, transform=self.random_transforms, cache_rate=1.0)[0] + self.sub_epochs = cfg.val_sub_epochs + self.len = len(self.monai_ds["image"]) + + def __getitem__(self, idx): + + if self.mode == "train": + monai_dict = self.monai_ds[idx // self.sub_epochs] + feature_dict = { + "input": torch.stack([item["image"] for item in monai_dict]), + "target": torch.stack([item["label"] for item in monai_dict]), + } + + else: + monai_dict = {k: self.monai_ds[k][idx] for k in self.monai_ds} + monai_dict["location"] = torch.from_numpy(self.monai_ds["image"].meta["location"][:, idx]) + feature_dict = { + "input": torch.stack([item["image"] for item in [monai_dict]]), + "location": torch.stack([item["location"] for item in [monai_dict]]), + "target": torch.stack([item["label"] for item in [monai_dict]]), + } + + return feature_dict + + def __len__(self): + return self.len + + def load_one(self, experiment_id): + + img_fp = f"{self.data_folder}{experiment_id}" + try: + with zarr.open(img_fp + "/VoxelSpacing10.000/denoised.zarr") as zf: + img = np.array(zf[0]).transpose(2, 1, 0) + # img = np.array(zarr.open(img_fp + '/VoxelSpacing10.000/denoised.zarr')[0]).transpose(2,1,0) + except Exception as e: + print(e) + + centers = self.exp_dict.get_group(experiment_id)[["x", "y", "z"]].values / 10 + classes = self.exp_dict.get_group(experiment_id)["particle_type"].map(self.class2id).values + mask = np.zeros((self.n_classes,) + img.shape[-3:]) + mask[classes, centers[:, 0].astype(int), centers[:, 1].astype(int), centers[:, 2].astype(int)] = 1 + return {"image": img, "label": mask} diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/metrics/metric_1.py b/competitions/kaggle/Cryo-ET/1st_place_solution/metrics/metric_1.py new file mode 100644 index 000000000..1d539d8ad --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/metrics/metric_1.py @@ -0,0 +1,201 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import numpy as np +import torch +from sklearn.metrics import roc_auc_score +import pandas as pd +from scipy.spatial import KDTree + + +class ParticipantVisibleError(Exception): + pass + + +def compute_metrics(reference_points, reference_radius, candidate_points): + num_reference_particles = len(reference_points) + num_candidate_particles = len(candidate_points) + + if len(reference_points) == 0: + return 0, num_candidate_particles, 0 + + if len(candidate_points) == 0: + return 0, 0, num_reference_particles + + ref_tree = KDTree(reference_points) + candidate_tree = KDTree(candidate_points) + raw_matches = candidate_tree.query_ball_tree(ref_tree, r=reference_radius) + matches_within_threshold = [] + for match in raw_matches: + matches_within_threshold.extend(match) + # Prevent submitting multiple matches per particle. + # This won't be be strictly correct in the (extremely rare) case where true particles + # are very close to each other. + matches_within_threshold = set(matches_within_threshold) + tp = int(len(matches_within_threshold)) + fp = int(num_candidate_particles - tp) + fn = int(num_reference_particles - tp) + return tp, fp, fn + + +def score( + solution: pd.DataFrame, + submission: pd.DataFrame, + row_id_column_name: str, + distance_multiplier: float, + beta: int, + weighted=True, +) -> float: + """ + F_beta + - a true positive occurs when + - (a) the predicted location is within a threshold of the particle radius, and + - (b) the correct `particle_type` is specified + - raw results (TP, FP, FN) are aggregated across all experiments for each particle type + - f_beta is calculated for each particle type + - individual f_beta scores are weighted by particle type for final score + """ + + particle_radius = { + "apo-ferritin": 60, + "beta-amylase": 65, + "beta-galactosidase": 90, + "ribosome": 150, + "thyroglobulin": 130, + "virus-like-particle": 135, + } + + weights = { + "apo-ferritin": 1, + "beta-amylase": 0, + "beta-galactosidase": 2, + "ribosome": 1, + "thyroglobulin": 2, + "virus-like-particle": 1, + } + + particle_radius = {k: v * distance_multiplier for k, v in particle_radius.items()} + + # Filter submission to only contain experiments found in the solution split + split_experiments = set(solution["experiment"].unique()) + submission = submission.loc[submission["experiment"].isin(split_experiments)] + + # Only allow known particle types + if not set(submission["particle_type"].unique()).issubset(set(weights.keys())): + raise ParticipantVisibleError("Unrecognized `particle_type`.") + + assert solution.duplicated(subset=["experiment", "x", "y", "z"]).sum() == 0 + assert particle_radius.keys() == weights.keys() + + results = {} + for particle_type in solution["particle_type"].unique(): + results[particle_type] = { + "total_tp": 0, + "total_fp": 0, + "total_fn": 0, + } + + for experiment in split_experiments: + for particle_type in solution["particle_type"].unique(): + reference_radius = particle_radius[particle_type] + select = (solution["experiment"] == experiment) & (solution["particle_type"] == particle_type) + reference_points = solution.loc[select, ["x", "y", "z"]].values + + select = (submission["experiment"] == experiment) & (submission["particle_type"] == particle_type) + candidate_points = submission.loc[select, ["x", "y", "z"]].values + + if len(reference_points) == 0: + reference_points = np.array([]) + reference_radius = 1 + + if len(candidate_points) == 0: + candidate_points = np.array([]) + + tp, fp, fn = compute_metrics(reference_points, reference_radius, candidate_points) + + results[particle_type]["total_tp"] += tp + results[particle_type]["total_fp"] += fp + results[particle_type]["total_fn"] += fn + + fbetas = [] + fbeta_weights = [] + particle_types = [] + for particle_type, totals in results.items(): + tp = totals["total_tp"] + fp = totals["total_fp"] + fn = totals["total_fn"] + + precision = tp / (tp + fp) if tp + fp > 0 else 0 + recall = tp / (tp + fn) if tp + fn > 0 else 0 + fbeta = ( + (1 + beta**2) * (precision * recall) / (beta**2 * precision + recall) if (precision + recall) > 0 else 0.0 + ) + fbetas += [fbeta] + fbeta_weights += [weights.get(particle_type, 1.0)] + particle_types += [particle_type] + + if weighted: + aggregate_fbeta = np.average(fbetas, weights=fbeta_weights) + else: + aggregate_fbeta = np.mean(fbetas) + + return aggregate_fbeta, dict(zip(particle_types, fbetas)) + + +def calc_metric(cfg, pp_out, val_df, pre="val"): + + particles = cfg.classes + pred_df = pp_out + + solution = val_df.copy() + solution["id"] = range(len(solution)) + + submission = pred_df.copy() + submission["experiment"] = solution["experiment"].unique()[0] + submission["id"] = range(len(submission)) + + best_ths = [] + for p in particles: + sol0a = solution[solution["particle_type"] == p].copy() + sub0a = submission[submission["particle_type"] == p].copy() + scores = [] + ths = np.arange(0, 0.5, 0.005) + for c in ths: + scores += [ + score( + sol0a.copy(), + sub0a[sub0a["conf"] > c].copy(), + row_id_column_name="id", + distance_multiplier=0.5, + beta=4, + weighted=False, + )[0] + ] + best_th = ths[np.argmax(scores)] + best_ths += [best_th] + + submission_pp = [] + for th, p in zip(best_ths, particles): + submission_pp += [submission[(submission["particle_type"] == p) & (submission["conf"] > th)].copy()] + submission_pp = pd.concat(submission_pp) + + score_pp, particle_scores = score( + solution[solution["particle_type"] != "beta-amylase"].copy(), + submission_pp.copy(), + row_id_column_name="id", + distance_multiplier=0.5, + beta=4, + ) + + result = {"score_" + k: v for k, v in particle_scores.items()} + result["score"] = score_pp + # print(result) + return result diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/models/mdl_1.py b/competitions/kaggle/Cryo-ET/1st_place_solution/models/mdl_1.py new file mode 100644 index 000000000..b8c871d4a --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/models/mdl_1.py @@ -0,0 +1,327 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import torch +import torch.nn.functional as F +from torch import nn +from torch.nn.parameter import Parameter +from torch.distributions import Beta +from monai.networks.nets.flexible_unet import SegmentationHead, UNetDecoder, FLEXUNET_BACKBONE + + +class PatchedUNetDecoder(UNetDecoder): + """add functionality to output all feature maps""" + + def forward(self, features: list[torch.Tensor], skip_connect: int = 4): + skips = features[:-1][::-1] + features = features[1:][::-1] + + out = [] + x = features[0] + out += [x] + for i, block in enumerate(self.blocks): + if i < skip_connect: + skip = skips[i] + else: + skip = None + x = block(x, skip) + out += [x] + return out + + +class FlexibleUNet(nn.Module): + """ + A flexible implementation of UNet-like encoder-decoder architecture. + + (Adjusted to support PatchDecoder and multi segmentation heads) + """ + + def __init__( + self, + in_channels: int, + out_channels: int, + backbone: str, + pretrained: bool = False, + decoder_channels: tuple = (256, 128, 64, 32, 16), + spatial_dims: int = 2, + norm: str | tuple = ("batch", {"eps": 1e-3, "momentum": 0.1}), + act: str | tuple = ("relu", {"inplace": True}), + dropout: float | tuple = 0.0, + decoder_bias: bool = False, + upsample: str = "nontrainable", + pre_conv: str = "default", + interp_mode: str = "nearest", + is_pad: bool = True, + ) -> None: + """ + A flexible implement of UNet, in which the backbone/encoder can be replaced with + any efficient or residual network. Currently the input must have a 2 or 3 spatial dimension + and the spatial size of each dimension must be a multiple of 32 if is_pad parameter + is False. + Please notice each output of backbone must be 2x downsample in spatial dimension + of last output. For example, if given a 512x256 2D image and a backbone with 4 outputs. + Spatial size of each encoder output should be 256x128, 128x64, 64x32 and 32x16. + + Args: + in_channels: number of input channels. + out_channels: number of output channels. + backbone: name of backbones to initialize, only support efficientnet and resnet right now, + can be from [efficientnet-b0, ..., efficientnet-b8, efficientnet-l2, resnet10, ..., resnet200]. + pretrained: whether to initialize pretrained weights. ImageNet weights are available for efficient networks + if spatial_dims=2 and batch norm is used. MedicalNet weights are available for residual networks + if spatial_dims=3 and in_channels=1. Default to False. + decoder_channels: number of output channels for all feature maps in decoder. + `len(decoder_channels)` should equal to `len(encoder_channels) - 1`,default + to (256, 128, 64, 32, 16). + spatial_dims: number of spatial dimensions, default to 2. + norm: normalization type and arguments, default to ("batch", {"eps": 1e-3, + "momentum": 0.1}). + act: activation type and arguments, default to ("relu", {"inplace": True}). + dropout: dropout ratio, default to 0.0. + decoder_bias: whether to have a bias term in decoder's convolution blocks. + upsample: upsampling mode, available options are``"deconv"``, ``"pixelshuffle"``, + ``"nontrainable"``. + pre_conv:a conv block applied before upsampling. Only used in the "nontrainable" or + "pixelshuffle" mode, default to `default`. + interp_mode: {``"nearest"``, ``"linear"``, ``"bilinear"``, ``"bicubic"``, ``"trilinear"``} + Only used in the "nontrainable" mode. + is_pad: whether to pad upsampling features to fit features from encoder. Default to True. + If this parameter is set to "True", the spatial dim of network input can be arbitrary + size, which is not supported by TensorRT. Otherwise, it must be a multiple of 32. + """ + super().__init__() + + if backbone not in FLEXUNET_BACKBONE.register_dict: + raise ValueError( + f"invalid model_name {backbone} found, must be one of {FLEXUNET_BACKBONE.register_dict.keys()}." + ) + + if spatial_dims not in (2, 3): + raise ValueError("spatial_dims can only be 2 or 3.") + + encoder = FLEXUNET_BACKBONE.register_dict[backbone] + self.backbone = backbone + self.spatial_dims = spatial_dims + encoder_parameters = encoder["parameter"] + if not ( + ("spatial_dims" in encoder_parameters) + and ("in_channels" in encoder_parameters) + and ("pretrained" in encoder_parameters) + ): + raise ValueError("The backbone init method must have spatial_dims, in_channels and pretrained parameters.") + encoder_feature_num = encoder["feature_number"] + if encoder_feature_num > 5: + raise ValueError("Flexible unet can only accept no more than 5 encoder feature maps.") + + decoder_channels = decoder_channels[:encoder_feature_num] + self.skip_connect = encoder_feature_num - 1 + encoder_parameters.update({"spatial_dims": spatial_dims, "in_channels": in_channels, "pretrained": pretrained}) + encoder_channels = tuple([in_channels] + list(encoder["feature_channel"])) + encoder_type = encoder["type"] + self.encoder = encoder_type(**encoder_parameters) + print(decoder_channels) + + self.decoder = PatchedUNetDecoder( + spatial_dims=spatial_dims, + encoder_channels=encoder_channels, + decoder_channels=decoder_channels, + act=act, + norm=norm, + dropout=dropout, + bias=decoder_bias, + upsample=upsample, + interp_mode=interp_mode, + pre_conv=pre_conv, + align_corners=None, + is_pad=is_pad, + ) + self.segmentation_heads = nn.ModuleList( + [ + SegmentationHead( + spatial_dims=spatial_dims, + in_channels=decoder_channel, + out_channels=out_channels + 1, + kernel_size=3, + act=None, + ) + for decoder_channel in decoder_channels[:-1] + ] + ) + + def forward(self, inputs: torch.Tensor): + + x = inputs + enc_out = self.encoder(x) + decoder_out = self.decoder(enc_out, self.skip_connect)[1:-1] + x_seg = [self.segmentation_heads[i](decoder_out[i]) for i in range(len(decoder_out))] + + return x_seg + + +def count_parameters(model): + return sum(p.numel() for p in model.parameters() if p.requires_grad) + + +def human_format(num): + num = float("{:.3g}".format(num)) + magnitude = 0 + while abs(num) >= 1000: + magnitude += 1 + num /= 1000.0 + return "{}{}".format("{:f}".format(num).rstrip("0").rstrip("."), ["", "K", "M", "B", "T"][magnitude]) + + +class DenseCrossEntropy(nn.Module): + def __init__(self, class_weights=None): + super(DenseCrossEntropy, self).__init__() + + self.class_weights = class_weights + + def forward(self, x, target): + x = x.float() + target = target.float() + logprobs = torch.nn.functional.log_softmax(x, dim=1, dtype=torch.float) + + loss = -logprobs * target + + class_losses = loss.mean((0, 2, 3, 4)) + if self.class_weights is not None: + loss = (class_losses * self.class_weights.to(class_losses.device)).sum() # / class_weights.sum() + else: + + loss = class_losses.sum() + return loss, class_losses + + +class Mixup(nn.Module): + def __init__(self, mix_beta, mixadd=False): + + super(Mixup, self).__init__() + self.beta_distribution = Beta(mix_beta, mix_beta) + self.mixadd = mixadd + + def forward(self, X, Y, Z=None): + + bs = X.shape[0] + n_dims = len(X.shape) + perm = torch.randperm(bs) + coeffs = self.beta_distribution.rsample(torch.Size((bs,))).to(X.device) + X_coeffs = coeffs.view((-1,) + (1,) * (X.ndim - 1)) + Y_coeffs = coeffs.view((-1,) + (1,) * (Y.ndim - 1)) + + X = X_coeffs * X + (1 - X_coeffs) * X[perm] + + if self.mixadd: + Y = (Y + Y[perm]).clip(0, 1) + else: + Y = Y_coeffs * Y + (1 - Y_coeffs) * Y[perm] + + if Z: + return X, Y, Z + + return X, Y + + +def to_ce_target(y): + # bs, c, h, w, d + y_bg = 1 - y.sum(1, keepdim=True).clamp(0, 1) + y = torch.cat([y, y_bg], 1) + y = y / y.sum(1, keepdim=True) + return y + + +class Net(nn.Module): + + def __init__(self, cfg): + super(Net, self).__init__() + + self.cfg = cfg + self.n_classes = cfg.n_classes + self.classes = cfg.classes + + self.backbone = FlexibleUNet(**cfg.backbone_args) + + self.mixup = Mixup(cfg.mixup_beta) + + print(f"Net parameters: {human_format(count_parameters(self))}") + self.lvl_weights = torch.from_numpy(cfg.lvl_weights) + self.loss_fn = DenseCrossEntropy(class_weights=torch.from_numpy(cfg.class_weights)) + + def forward(self, batch): + + x = batch["input"] + if "target" in batch.keys(): + y = batch["target"] + if self.training: + if torch.rand(1)[0] < self.cfg.mixup_p: + x, y = self.mixup(x, y) + out = self.backbone(x) + + outputs = {} + + if "target" in batch.keys(): + ys = [F.adaptive_max_pool3d(y, item.shape[-3:]) for item in out] + losses = torch.stack([self.loss_fn(out[i], to_ce_target(ys[i]))[0] for i in range(len(out))]) + lvl_weights = self.lvl_weights.to(losses.device) + loss = (losses * lvl_weights).sum() / lvl_weights.sum() + outputs["loss"] = loss + if not self.training: + outputs["logits"] = out[-1] + if "location" in batch: + outputs["location"] = batch["location"] + + return outputs + + +class TestNet(nn.Module): + + def __init__(self, **backbone_args): + super(TestNet, self).__init__() + + self.backbone = FlexibleUNet(**backbone_args) + + def forward(self, x): + # x shape is bs, c, h, w, d + out = self.backbone(x) + # out shape is bs, 7, h//2, w//2, d//2 + logits = out[-1] # for heatmap do softmax + reorder classes .softmax(1)[:,[0,2,3,4,5,1]] + return logits + + +# import torch +# import monai.transforms as mt +# import zarr +# import numpy as np + +# def preprocess_img(zarr_fn, transforms): +# img = np.array(zarr.open(zarr_fn)[0]).transpose(2,1,0) +# img = transforms({'image':img})['image'] +# return img + +# backbone_args = dict(spatial_dims=3, +# in_channels=1, +# out_channels=6, +# backbone='resnet34', +# pretrained=False) + +# net = TestNet(**backbone_args) +# sd = torch.load(CKPT_FILE)['model'] +# net.eval().cuda() +# net.load_state_dict(sd) + +# static_transforms = mt.Compose([mt.EnsureChannelFirstd(keys=["image"], channel_dim="no_channel"),mt.NormalizeIntensityd(keys="image"),]) +# zarr_fn = '/mount/cryo/data/czii-cryo-et-object-identification/train/static/ExperimentRuns/TS_5_4/VoxelSpacing10.000/denoised.zarr' +# img = preprocess_img(zarr_fn,static_transforms) # torch.Size([1, 630, 630, 184]) + +# patch = img[None, :, :96, :96, :96] # torch.Size([1, ,1 96, 96, 96]) + +# logits = net(patch.cuda()) +# proba_heatmap = logits.softmax(1)[:,[0,2,3,4,5,1]] diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/partly_Unet.png b/competitions/kaggle/Cryo-ET/1st_place_solution/partly_Unet.png new file mode 100644 index 000000000..7f4f72cdb Binary files /dev/null and b/competitions/kaggle/Cryo-ET/1st_place_solution/partly_Unet.png differ diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/postprocess/pp_1.py b/competitions/kaggle/Cryo-ET/1st_place_solution/postprocess/pp_1.py new file mode 100644 index 000000000..d0dee3e35 --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/postprocess/pp_1.py @@ -0,0 +1,78 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import pandas as pd +import torch +from torch.nn import functional as F +from tqdm import tqdm + +import torch +from torch import nn + + +def simple_nms(scores, nms_radius: int): + """Fast Non-maximum suppression to remove nearby points""" + assert nms_radius >= 0 + + def max_pool(x): + return torch.nn.functional.max_pool3d(x, kernel_size=nms_radius * 2 + 1, stride=1, padding=nms_radius) + + zeros = torch.zeros_like(scores) + max_mask = scores == max_pool(scores) + return torch.where(max_mask, scores, zeros) + + +def reconstruct(img, locations, out_size, crop_size): + reconstructed_img = torch.zeros(out_size) + + for i in range(img.shape[0]): + reconstructed_img[ + :, + locations[0][i] : locations[0][i] + crop_size[0], + locations[1][i] : locations[1][i] + crop_size[1], + locations[2][i] : locations[2][i] + crop_size[2], + ] = img[i, :] + return reconstructed_img + + +def post_process_pipeline(cfg, val_data, val_df): + + img = val_data["logits"] + img = torch.nn.functional.interpolate( + img, size=(cfg.roi_size[0], cfg.roi_size[1], cfg.roi_size[2]), mode="trilinear", align_corners=False + ) + locations = val_data["location"] + out_size = [cfg.n_classes + 1] + [l.item() + r for l, r in zip(locations.max(0)[0], cfg.roi_size)] + rec_img = reconstruct(img, locations.permute(1, 0), out_size=out_size, crop_size=cfg.roi_size) + s = rec_img.shape[-3:] + rec_img = torch.nn.functional.interpolate( + rec_img[None], size=(s[0] // 2, s[1] // 2, s[2] // 2), mode="trilinear", align_corners=False + )[0] + preds = rec_img.softmax(0)[:-1] + + pred_df = [] + + for i, p in enumerate(cfg.classes): + p1 = preds[i][None,].cuda() + y = simple_nms(p1, nms_radius=int(0.5 * cfg.particle_radi[p] / 10)) + kps = torch.where(y > 0) + xyz = torch.stack(kps[1:], -1) * 10 * 2 + conf = y[kps] + pred_df_ = pd.DataFrame(xyz.cpu().numpy(), columns=["x", "y", "z"]) + pred_df_["particle_type"] = p + pred_df_["conf"] = conf.cpu().numpy() + pred_df += [pred_df_] + pred_df = pd.concat(pred_df) + pred_df = pred_df[ + (pred_df["x"] < 6300) & (pred_df["y"] < 6300) & (pred_df["z"] < 1840) & (pred_df["conf"] > 0.01) + ].copy() + pred_df.to_csv(f"{cfg.output_dir}/fold{cfg.fold}/val_pred_df_seed{cfg.seed}.csv", index=False) + return pred_df diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/requirements.txt b/competitions/kaggle/Cryo-ET/1st_place_solution/requirements.txt new file mode 100644 index 000000000..fcc88d67c --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/requirements.txt @@ -0,0 +1,3 @@ +zarr +monai==1.4.0 +kaggle==1.7.4.2 diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/train.py b/competitions/kaggle/Cryo-ET/1st_place_solution/train.py new file mode 100644 index 000000000..c2e3589e2 --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/train.py @@ -0,0 +1,438 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +import numpy as np +import pandas as pd +import importlib +import sys +from tqdm import tqdm +import gc +import argparse +import torch +import math + +try: + from torch.amp import GradScaler, autocast +except: + from torch.cuda.amp import GradScaler, autocast +from torch.nn.parallel import DistributedDataParallel as NativeDDP +from collections import defaultdict + +from utils import ( + sync_across_gpus, + set_seed, + get_model, + create_checkpoint, + load_checkpoint, + get_data, + get_dataset, + get_dataloader, + calc_grad_norm, + calc_weight_norm, +) +from utils import ( + get_optimizer, + get_scheduler, +) + + +from copy import copy +import os + +os.environ["OMP_NUM_THREADS"] = "1" +os.environ["MKL_NUM_THREADS"] = "1" +os.environ["OPENBLAS_NUM_THREADS"] = "1" +os.environ["VECLIB_MAXIMUM_THREADS"] = "1" +os.environ["NUMEXPR_NUM_THREADS"] = "1" +os.environ["TOKENIZERS_PARALLELISM"] = "false" + +try: + import cv2 + + cv2.setNumThreads(0) +except: + print("no cv2 installed, running without") + + +sys.path.append("configs") +sys.path.append("models") +sys.path.append("data") +sys.path.append("postprocess") +sys.path.append("metrics") + + +def run_eval(model, val_dataloader, cfg, pre="val", curr_epoch=0): + saved_images = False + model.eval() + torch.set_grad_enabled(False) + + # store information for evaluation + val_data = defaultdict(list) + val_score = 0 + for ind_, data in enumerate(tqdm(val_dataloader, disable=(cfg.local_rank != 0) | cfg.disable_tqdm)): + + batch = cfg.batch_to_device(data, cfg.device) + + if cfg.mixed_precision: + with autocast("cuda"): + output = model(batch) + else: + output = model(batch) + + if (cfg.local_rank == 0) and (cfg.calc_metric) and (((curr_epoch + 1) % cfg.calc_metric_epochs) == 0): + # per batch calculations + pass + + if (not saved_images) & (cfg.save_first_batch_preds): + save_first_batch_preds(batch, output, cfg) + saved_images = True + + for key, val in output.items(): + val_data[key] += [output[key]] + + for key, val in output.items(): + value = val_data[key] + if isinstance(value[0], list): + val_data[key] = [item for sublist in value for item in sublist] + + else: + if len(value[0].shape) == 0: + val_data[key] = torch.stack(value) + else: + val_data[key] = torch.cat(value, dim=0) + + if (cfg.local_rank == 0) and (cfg.calc_metric) and (((curr_epoch + 1) % cfg.calc_metric_epochs) == 0): + pass + + if cfg.distributed and cfg.eval_ddp: + for key, val in output.items(): + val_data[key] = sync_across_gpus(val_data[key], cfg.world_size) + + if cfg.local_rank == 0: + if cfg.save_val_data: + if cfg.distributed: + for k, v in val_data.items(): + val_data[k] = v[: len(val_dataloader.dataset)] + torch.save(val_data, f"{cfg.output_dir}/fold{cfg.fold}/{pre}_data_seed{cfg.seed}.pth") + + loss_names = [key for key in output if "loss" in key] + loss_names += [key for key in output if "score" in key] + for k in loss_names: + if cfg.local_rank == 0 and k in val_data: + losses = val_data[k].cpu().numpy() + loss = np.mean(losses) + + print(f"Mean {pre}_{k}", loss) + + if (cfg.local_rank == 0) and (cfg.calc_metric) and (((curr_epoch + 1) % cfg.calc_metric_epochs) == 0): + + val_df = val_dataloader.dataset.df + pp_out = cfg.post_process_pipeline(cfg, val_data, val_df) + val_score = cfg.calc_metric(cfg, pp_out, val_df, pre) + if type(val_score) != dict: + val_score = {f"score": val_score} + + for k, v in val_score.items(): + print(f"{pre}_{k}: {v:.3f}") + + if cfg.distributed: + torch.distributed.barrier() + + # print("EVAL FINISHED") + + return val_score + + +def train(cfg): + # set seed + if cfg.seed < 0: + cfg.seed = np.random.randint(1_000_000) + print("seed", cfg.seed) + + if cfg.distributed: + + cfg.local_rank = int(os.environ["LOCAL_RANK"]) + device = "cuda:%d" % cfg.local_rank + cfg.device = device + + torch.cuda.set_device(cfg.local_rank) + torch.distributed.init_process_group(backend="nccl", init_method="env://") + cfg.world_size = torch.distributed.get_world_size() + cfg.rank = torch.distributed.get_rank() + # print("Training in distributed mode with multiple processes, 1 GPU per process.") + print(f"Process {cfg.rank}, total {cfg.world_size}, local rank {cfg.local_rank}.") + cfg.group = torch.distributed.new_group(np.arange(cfg.world_size)) + # print("Group", cfg.group) + + # syncing the random seed + cfg.seed = int( + sync_across_gpus(torch.Tensor([cfg.seed]).to(device), cfg.world_size).detach().cpu().numpy()[0] + ) # + + print(f"LOCAL_RANK {cfg.local_rank}, device {device}, seed {cfg.seed}") + + else: + cfg.local_rank = 0 + cfg.world_size = 1 + cfg.rank = 0 # global rank + + device = "cuda:%d" % cfg.gpu + cfg.device = device + + set_seed(cfg.seed) + + train_df, val_df, test_df = get_data(cfg) + + train_dataset = get_dataset(train_df, cfg, mode="train") + train_dataloader = get_dataloader(train_dataset, cfg, mode="train") + + val_dataset = get_dataset(val_df, cfg, mode="val") + val_dataloader = get_dataloader(val_dataset, cfg, mode="val") + + if cfg.test: + test_dataset = get_dataset(test_df, cfg, mode="test") + test_dataloader = get_dataloader(test_dataset, cfg, mode="test") + + if cfg.train_val: + train_val_dataset = get_dataset(train_df, cfg, mode="val") + train_val_dataloader = get_dataloader(train_val_dataset, cfg, "val") + + model = get_model(cfg, train_dataset) + if cfg.compile_model: + print("compiling model") + model = torch.compile(model) + model.to(device) + + if cfg.distributed: + + if cfg.syncbn: + model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model) + + model = NativeDDP(model, device_ids=[cfg.local_rank], find_unused_parameters=cfg.find_unused_parameters) + + total_steps = len(train_dataset) + if train_dataloader.sampler is not None: + if "WeightedRandomSampler" in str(train_dataloader.sampler.__class__): + total_steps = train_dataloader.sampler.num_samples + + optimizer = get_optimizer(model, cfg) + scheduler = get_scheduler(cfg, optimizer, total_steps) + + if cfg.mixed_precision: + scaler = GradScaler() + else: + scaler = None + + cfg.curr_step = 0 + i = 0 + best_val_loss = np.inf + optimizer.zero_grad() + total_grad_norm = None + total_weight_norm = None + total_grad_norm_after_clip = None + + for epoch in range(cfg.epochs): + + set_seed(cfg.seed + epoch + cfg.local_rank) + + cfg.curr_epoch = epoch + if cfg.local_rank == 0: + print("EPOCH:", epoch) + + if cfg.distributed: + train_dataloader.sampler.set_epoch(epoch) + + progress_bar = tqdm(range(len(train_dataloader)), disable=cfg.disable_tqdm) + tr_it = iter(train_dataloader) + + losses = [] + + gc.collect() + + if cfg.train: + # ==== TRAIN LOOP + for itr in progress_bar: + i += 1 + + cfg.curr_step += cfg.batch_size * cfg.world_size + + try: + data = next(tr_it) + except Exception as e: + print(e) + print("DATA FETCH ERROR") + # continue + + model.train() + torch.set_grad_enabled(True) + + batch = cfg.batch_to_device(data, device) + + if cfg.mixed_precision: + with autocast("cuda"): + output_dict = model(batch) + else: + if cfg.bf16: + with autocast("cuda", dtype=torch.bfloat16): + output_dict = model(batch) + else: + output_dict = model(batch) + + loss = output_dict["loss"] + + losses.append(loss.item()) + + if cfg.grad_accumulation > 1: + loss /= cfg.grad_accumulation + + # Backward pass + + if cfg.mixed_precision: + scaler.scale(loss).backward() + + if i % cfg.grad_accumulation == 0: + if (cfg.track_grad_norm) or (cfg.clip_grad > 0): + scaler.unscale_(optimizer) + if cfg.track_grad_norm: + total_grad_norm = calc_grad_norm(model.parameters(), cfg.grad_norm_type) + if cfg.clip_grad > 0: + torch.nn.utils.clip_grad_norm_(model.parameters(), cfg.clip_grad) + if cfg.track_grad_norm: + total_grad_norm_after_clip = calc_grad_norm(model.parameters(), cfg.grad_norm_type) + scaler.step(optimizer) + scaler.update() + optimizer.zero_grad() + + else: + + loss.backward() + if i % cfg.grad_accumulation == 0: + if cfg.track_grad_norm: + total_grad_norm = calc_grad_norm(model.parameters()) + if cfg.clip_grad > 0: + torch.nn.utils.clip_grad_norm_(model.parameters(), cfg.clip_grad) + if cfg.track_grad_norm: + total_grad_norm_after_clip = calc_grad_norm(model.parameters(), cfg.grad_norm_type) + if cfg.track_weight_norm: + total_weight_norm = calc_weight_norm(model.parameters(), cfg.grad_norm_type) + optimizer.step() + optimizer.zero_grad() + # print(optimizer.state_dict()) + # break + + if cfg.distributed: + torch.cuda.synchronize() + + if scheduler is not None: + scheduler.step() + + if cfg.local_rank == 0 and cfg.curr_step % cfg.batch_size == 0: + + loss_names = [key for key in output_dict if "loss" in key] + + progress_bar.set_description(f"loss: {np.mean(losses[-10:]):.4f}") + + if cfg.eval_steps != 0: + if i % cfg.eval_steps == 0: + if cfg.distributed and cfg.eval_ddp: + val_loss = run_eval(model, val_dataloader, cfg, pre="val", curr_epoch=epoch) + else: + if cfg.local_rank == 0: + val_loss = run_eval(model, val_dataloader, cfg, pre="val", curr_epoch=epoch) + else: + val_score = 0 + + print(f"Mean train_loss {np.mean(losses):.4f}") + + if cfg.distributed: + torch.cuda.synchronize() + if cfg.force_fp16: + model = model.half().float() + if cfg.val: + + if (epoch + 1) % cfg.eval_epochs == 0 or (epoch + 1) == cfg.epochs: + if cfg.distributed and cfg.eval_ddp: + val_score = run_eval(model, val_dataloader, cfg, pre="val", curr_epoch=epoch) + else: + if cfg.local_rank == 0: + val_score = run_eval(model, val_dataloader, cfg, pre="val", curr_epoch=epoch) + else: + val_score = 0 + + if cfg.train_val == True: + if (epoch + 1) % cfg.eval_train_epochs == 0 or (epoch + 1) == cfg.epochs: + if cfg.distributed and cfg.eval_ddp: + _ = get_preds(model, train_val_dataloader, cfg, pre=cfg.pre_train_val) + + else: + if cfg.local_rank == 0: + _ = get_preds(model, train_val_dataloader, cfg, pre=cfg.pre_train_val) + + if cfg.distributed: + torch.distributed.barrier() + + if (cfg.local_rank == 0) and (cfg.epochs > 0) and (cfg.save_checkpoint): + if not cfg.save_only_last_ckpt: + checkpoint = create_checkpoint(cfg, model, optimizer, epoch, scheduler=scheduler, scaler=scaler) + + torch.save(checkpoint, f"{cfg.output_dir}/fold{cfg.fold}/checkpoint_last_seed{cfg.seed}.pth") + + if (cfg.local_rank == 0) and (cfg.epochs > 0) and (cfg.save_checkpoint): + checkpoint = create_checkpoint(cfg, model, optimizer, epoch, scheduler=scheduler, scaler=scaler) + + torch.save(checkpoint, f"{cfg.output_dir}/fold{cfg.fold}/checkpoint_last_seed{cfg.seed}.pth") + + if cfg.test: + run_eval(model, test_dataloader, test_df, cfg, pre="test") + + return val_score + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser(description="") + + parser.add_argument("-C", "--config", help="config filename") + parser_args, other_args = parser.parse_known_args(sys.argv) + + cfg = copy(importlib.import_module(parser_args.config).cfg) + + # overwrite params in config with additional args + if len(other_args) > 1: + other_args = {k.replace("-", ""): v for k, v in zip(other_args[1::2], other_args[2::2])} + + for key in other_args: + if key in cfg.__dict__: + + print(f"overwriting cfg.{key}: {cfg.__dict__[key]} -> {other_args[key]}") + cfg_type = type(cfg.__dict__[key]) + if other_args[key] == "None": + cfg.__dict__[key] = None + elif cfg_type == bool: + cfg.__dict__[key] = other_args[key] == "True" + elif cfg_type == type(None): + cfg.__dict__[key] = other_args[key] + else: + cfg.__dict__[key] = cfg_type(other_args[key]) + + os.makedirs(str(cfg.output_dir + f"/fold{cfg.fold}/"), exist_ok=True) + + cfg.CustomDataset = importlib.import_module(cfg.dataset).CustomDataset + cfg.tr_collate_fn = importlib.import_module(cfg.dataset).tr_collate_fn + cfg.val_collate_fn = importlib.import_module(cfg.dataset).val_collate_fn + cfg.batch_to_device = importlib.import_module(cfg.dataset).batch_to_device + + cfg.post_process_pipeline = importlib.import_module(cfg.post_process_pipeline).post_process_pipeline + cfg.calc_metric = importlib.import_module(cfg.metric).calc_metric + + result = train(cfg) + print(result) diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/train_folded_v1.csv b/competitions/kaggle/Cryo-ET/1st_place_solution/train_folded_v1.csv new file mode 100644 index 000000000..3b4c53669 --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/train_folded_v1.csv @@ -0,0 +1,1270 @@ +x,y,z,particle_type,experiment,fold +5417.091,1120.574,567.527,beta-amylase,TS_5_4,0 +4340.0,1220.0,365.0,beta-amylase,TS_5_4,0 +4908.973,1745.891,279.668,beta-amylase,TS_5_4,0 +5153.2,4595.2,420.8,beta-amylase,TS_5_4,0 +1500.0,4212.222,1016.667,beta-amylase,TS_5_4,0 +1068.16,3620.736,1033.129,beta-amylase,TS_5_4,0 +1025.185,2153.333,1215.556,beta-amylase,TS_5_4,0 +1750.0,2648.75,881.25,beta-amylase,TS_5_4,0 +2609.328,2513.109,1131.555,beta-amylase,TS_5_4,0 +3634.655,3226.691,1280.655,beta-amylase,TS_5_4,0 +1959.181,701.502,371.229,beta-galactosidase,TS_5_4,0 +2135.713,4459.195,936.655,beta-galactosidase,TS_5_4,0 +3449.782,5052.93,511.041,beta-galactosidase,TS_5_4,0 +4623.597,3652.095,1161.66,beta-galactosidase,TS_5_4,0 +1390.238,4121.667,570.595,beta-galactosidase,TS_5_4,0 +1317.617,2292.128,1089.064,beta-galactosidase,TS_5_4,0 +3540.625,1851.25,50.0,beta-galactosidase,TS_5_4,0 +5421.651,1158.868,563.302,beta-galactosidase,TS_5_4,0 +5132.863,4594.658,436.41,beta-galactosidase,TS_5_4,0 +455.241,2838.414,623.402,beta-galactosidase,TS_5_4,0 +766.848,1452.913,700.413,beta-galactosidase,TS_5_4,0 +4048.927,5220.881,1249.54,beta-galactosidase,TS_5_4,0 +4601.271,601.066,600.934,ribosome,TS_5_4,0 +4803.789,455.425,514.016,ribosome,TS_5_4,0 +4715.436,825.374,802.166,ribosome,TS_5_4,0 +5003.275,782.745,802.48,ribosome,TS_5_4,0 +710.459,3815.845,1405.435,ribosome,TS_5_4,0 +2421.852,1141.637,663.19,ribosome,TS_5_4,0 +691.64,897.383,288.35,ribosome,TS_5_4,0 +948.219,2155.294,372.13,ribosome,TS_5_4,0 +1910.761,755.539,1052.241,ribosome,TS_5_4,0 +1610.634,232.123,1068.46,ribosome,TS_5_4,0 +4217.488,4192.518,941.839,ribosome,TS_5_4,0 +4725.193,4090.105,1110.456,ribosome,TS_5_4,0 +2545.405,3069.551,1119.704,ribosome,TS_5_4,0 +4091.497,2159.453,874.047,ribosome,TS_5_4,0 +3986.554,1885.673,744.371,ribosome,TS_5_4,0 +3822.144,1883.156,1000.826,ribosome,TS_5_4,0 +2803.199,2701.106,336.099,ribosome,TS_5_4,0 +3189.676,2565.629,306.801,ribosome,TS_5_4,0 +3526.519,2317.862,1059.979,ribosome,TS_5_4,0 +4632.091,2990.234,244.138,ribosome,TS_5_4,0 +4620.799,2802.16,513.43,ribosome,TS_5_4,0 +4809.673,2925.589,730.053,ribosome,TS_5_4,0 +5072.909,2902.993,763.206,ribosome,TS_5_4,0 +4790.666,2517.002,489.436,ribosome,TS_5_4,0 +4805.633,2619.184,874.657,ribosome,TS_5_4,0 +5056.98,2532.435,877.002,ribosome,TS_5_4,0 +3397.301,4660.778,1309.286,ribosome,TS_5_4,0 +2843.556,6150.71,818.754,ribosome,TS_5_4,0 +4195.391,3961.637,866.076,ribosome,TS_5_4,0 +4142.395,2205.03,1126.4,ribosome,TS_5_4,0 +4356.021,3945.454,1156.437,ribosome,TS_5_4,0 +4527.706,221.468,279.908,thyroglobulin,TS_5_4,0 +5458.275,3743.418,332.075,thyroglobulin,TS_5_4,0 +5086.266,5378.977,411.051,thyroglobulin,TS_5_4,0 +2091.242,489.208,433.742,thyroglobulin,TS_5_4,0 +4338.644,1822.464,455.595,thyroglobulin,TS_5_4,0 +1310.154,4513.909,442.042,thyroglobulin,TS_5_4,0 +1585.642,3326.048,517.486,thyroglobulin,TS_5_4,0 +2293.097,1869.88,555.296,thyroglobulin,TS_5_4,0 +3469.929,3753.49,782.101,thyroglobulin,TS_5_4,0 +2439.53,3199.849,797.353,thyroglobulin,TS_5_4,0 +3088.296,2349.548,839.11,thyroglobulin,TS_5_4,0 +756.786,3307.765,923.717,thyroglobulin,TS_5_4,0 +3095.628,5843.591,947.596,thyroglobulin,TS_5_4,0 +2273.228,1492.983,949.712,thyroglobulin,TS_5_4,0 +5121.456,5468.983,1005.202,thyroglobulin,TS_5_4,0 +474.398,3091.932,1076.289,thyroglobulin,TS_5_4,0 +1541.559,5808.044,1379.853,thyroglobulin,TS_5_4,0 +441.346,2266.635,835.865,thyroglobulin,TS_5_4,0 +873.75,2086.875,1055.312,thyroglobulin,TS_5_4,0 +2405.172,6025.0,563.966,thyroglobulin,TS_5_4,0 +1990.0,5790.0,870.0,thyroglobulin,TS_5_4,0 +3087.172,5274.276,700.759,thyroglobulin,TS_5_4,0 +2917.805,3958.293,730.732,thyroglobulin,TS_5_4,0 +3419.647,2830.627,459.333,thyroglobulin,TS_5_4,0 +5090.0,573.478,122.174,thyroglobulin,TS_5_4,0 +2175.491,3100.549,452.861,thyroglobulin,TS_5_4,0 +2335.068,4957.432,703.716,thyroglobulin,TS_5_4,0 +2752.287,5760.574,881.249,thyroglobulin,TS_5_4,0 +2613.583,2253.208,884.042,thyroglobulin,TS_5_4,0 +534.129,1666.452,1038.0,thyroglobulin,TS_5_4,0 +626.933,1151.091,637.139,virus-like-particle,TS_5_4,0 +1480.552,1278.98,820.227,virus-like-particle,TS_5_4,0 +1251.324,5451.893,735.631,virus-like-particle,TS_5_4,0 +117.204,5393.268,923.733,virus-like-particle,TS_5_4,0 +672.259,3039.65,711.322,virus-like-particle,TS_5_4,0 +2620.699,3741.685,865.897,virus-like-particle,TS_5_4,0 +2636.539,4214.98,965.41,virus-like-particle,TS_5_4,0 +3137.396,3572.46,372.914,virus-like-particle,TS_5_4,0 +3294.133,3027.464,674.07,virus-like-particle,TS_5_4,0 +2997.686,4948.218,1169.375,virus-like-particle,TS_5_4,0 +3435.851,6177.824,1016.473,virus-like-particle,TS_5_4,0 +468.514,5915.906,604.167,apo-ferritin,TS_5_4,0 +5674.694,1114.354,565.068,apo-ferritin,TS_5_4,0 +5744.509,1049.172,653.712,apo-ferritin,TS_5_4,0 +5880.769,1125.348,579.56,apo-ferritin,TS_5_4,0 +4661.667,1269.497,810.409,apo-ferritin,TS_5_4,0 +4593.498,1195.874,877.982,apo-ferritin,TS_5_4,0 +4569.735,1249.967,735.033,apo-ferritin,TS_5_4,0 +4524.726,1359.207,763.445,apo-ferritin,TS_5_4,0 +1072.449,2963.265,384.014,apo-ferritin,TS_5_4,0 +1068.182,3024.848,515.628,apo-ferritin,TS_5_4,0 +995.294,3029.412,407.899,apo-ferritin,TS_5_4,0 +969.749,3086.695,557.95,apo-ferritin,TS_5_4,0 +1088.28,3115.16,609.8,apo-ferritin,TS_5_4,0 +1130.0,3130.0,795.0,apo-ferritin,TS_5_4,0 +891.818,2977.273,510.606,apo-ferritin,TS_5_4,0 +924.121,3210.767,736.134,apo-ferritin,TS_5_4,0 +1663.767,3311.767,1051.7,apo-ferritin,TS_5_4,0 +5870.268,5131.104,76.321,apo-ferritin,TS_5_4,0 +4807.948,4949.03,1039.478,apo-ferritin,TS_5_4,0 +4837.967,5049.35,1030.081,apo-ferritin,TS_5_4,0 +4714.452,5064.976,1077.238,apo-ferritin,TS_5_4,0 +5062.45,5082.947,1205.927,apo-ferritin,TS_5_4,0 +4829.392,4473.384,208.707,apo-ferritin,TS_5_4,0 +3684.954,4537.92,639.235,apo-ferritin,TS_5_4,0 +3738.736,4124.944,1182.974,apo-ferritin,TS_5_4,0 +2682.13,3630.154,642.963,apo-ferritin,TS_5_4,0 +2801.725,1630.415,204.473,apo-ferritin,TS_5_4,0 +2504.12,2311.852,267.454,apo-ferritin,TS_5_4,0 +2499.282,2197.394,298.989,apo-ferritin,TS_5_4,0 +2626.828,2295.534,254.337,apo-ferritin,TS_5_4,0 +2552.162,2711.815,416.023,apo-ferritin,TS_5_4,0 +609.824,4421.479,678.38,apo-ferritin,TS_5_4,0 +521.952,4275.286,650.429,apo-ferritin,TS_5_4,0 +612.529,4271.092,494.828,apo-ferritin,TS_5_4,0 +4657.21,5130.873,989.265,apo-ferritin,TS_5_4,0 +1151.918,3210.74,696.075,apo-ferritin,TS_5_4,0 +1038.494,3201.114,695.662,apo-ferritin,TS_5_4,0 +1059.84,3172.472,836.249,apo-ferritin,TS_5_4,0 +1066.853,3207.492,585.938,apo-ferritin,TS_5_4,0 +2340.306,2283.227,345.429,apo-ferritin,TS_5_4,0 +2349.924,2329.502,485.604,apo-ferritin,TS_5_4,0 +3779.642,4198.01,1296.612,apo-ferritin,TS_5_4,0 +611.797,4393.753,515.641,apo-ferritin,TS_5_4,0 +557.665,4367.989,605.753,apo-ferritin,TS_5_4,0 +5715.014,4998.216,115.143,apo-ferritin,TS_5_4,0 +5748.253,5115.846,108.033,apo-ferritin,TS_5_4,0 +320.256,5482.051,1088.205,beta-amylase,TS_86_3,1 +4290.0,2290.0,1050.0,beta-amylase,TS_86_3,1 +4268.588,1190.235,706.118,beta-amylase,TS_86_3,1 +3693.165,1512.658,849.62,beta-amylase,TS_86_3,1 +4384.167,445.833,1119.167,beta-amylase,TS_86_3,1 +5437.751,1527.337,824.201,beta-amylase,TS_86_3,1 +5509.412,4910.588,701.569,beta-amylase,TS_86_3,1 +2247.754,378.116,1050.217,beta-amylase,TS_86_3,1 +1188.393,2327.798,696.845,beta-amylase,TS_86_3,1 +348.342,3368.995,520.0,beta-galactosidase,TS_86_3,1 +3111.559,1404.395,597.264,beta-galactosidase,TS_86_3,1 +5128.266,4932.229,585.512,beta-galactosidase,TS_86_3,1 +3984.278,423.056,725.944,beta-galactosidase,TS_86_3,1 +2763.251,3038.958,855.062,beta-galactosidase,TS_86_3,1 +4073.83,3599.172,1070.207,beta-galactosidase,TS_86_3,1 +4463.183,1681.903,1096.54,beta-galactosidase,TS_86_3,1 +1304.049,2964.441,1180.928,beta-galactosidase,TS_86_3,1 +5695.682,831.818,827.273,beta-galactosidase,TS_86_3,1 +2369.583,5805.208,949.167,beta-galactosidase,TS_86_3,1 +1808.132,5982.125,1006.52,beta-galactosidase,TS_86_3,1 +1549.167,3946.437,574.713,beta-galactosidase,TS_86_3,1 +4913.909,2197.738,878.413,beta-galactosidase,TS_86_3,1 +565.0,1044.148,1032.481,beta-galactosidase,TS_86_3,1 +1908.833,229.833,948.167,beta-galactosidase,TS_86_3,1 +2474.59,4055.164,488.224,beta-galactosidase,TS_86_3,1 +4085.02,1689.352,644.332,beta-galactosidase,TS_86_3,1 +2511.429,480.0,807.619,beta-galactosidase,TS_86_3,1 +4450.0,1356.667,876.667,beta-galactosidase,TS_86_3,1 +2000.686,1232.08,923.473,beta-galactosidase,TS_86_3,1 +3552.92,3973.993,1034.795,beta-galactosidase,TS_86_3,1 +5021.325,3944.498,1080.0,beta-galactosidase,TS_86_3,1 +2745.385,872.86,1044.339,beta-galactosidase,TS_86_3,1 +4662.299,5907.258,945.084,ribosome,TS_86_3,1 +4504.116,5758.297,1102.158,ribosome,TS_86_3,1 +4462.021,6207.162,999.208,ribosome,TS_86_3,1 +4777.299,6156.435,1063.849,ribosome,TS_86_3,1 +4666.205,5215.513,747.842,ribosome,TS_86_3,1 +4438.999,4935.4,788.27,ribosome,TS_86_3,1 +5252.162,4295.573,849.171,ribosome,TS_86_3,1 +5689.451,4321.102,893.643,ribosome,TS_86_3,1 +6095.703,4218.054,560.215,ribosome,TS_86_3,1 +4798.895,4427.67,644.16,ribosome,TS_86_3,1 +4790.113,4162.789,708.444,ribosome,TS_86_3,1 +4327.26,4594.434,562.929,ribosome,TS_86_3,1 +4530.23,4362.667,507.066,ribosome,TS_86_3,1 +4685.378,4744.749,747.362,ribosome,TS_86_3,1 +4546.785,4603.197,1000.327,ribosome,TS_86_3,1 +5815.688,5910.474,1112.865,ribosome,TS_86_3,1 +3558.463,2113.569,1179.779,ribosome,TS_86_3,1 +2024.391,3063.731,968.932,ribosome,TS_86_3,1 +1917.914,2905.379,1174.505,ribosome,TS_86_3,1 +1649.253,2893.382,1100.787,ribosome,TS_86_3,1 +1665.413,3181.988,1083.413,ribosome,TS_86_3,1 +1563.059,6022.91,504.966,ribosome,TS_86_3,1 +1217.9,3972.649,1086.719,ribosome,TS_86_3,1 +1352.382,4356.122,751.612,ribosome,TS_86_3,1 +1213.789,4280.218,1050.137,ribosome,TS_86_3,1 +1444.336,4663.022,1049.317,ribosome,TS_86_3,1 +1974.574,4645.358,1227.115,ribosome,TS_86_3,1 +1764.31,4357.96,1082.894,ribosome,TS_86_3,1 +1677.802,4682.915,728.72,ribosome,TS_86_3,1 +615.248,5884.541,961.487,ribosome,TS_86_3,1 +253.933,5845.176,596.784,ribosome,TS_86_3,1 +238.655,5198.089,1175.451,ribosome,TS_86_3,1 +571.25,4918.854,1019.915,ribosome,TS_86_3,1 +536.338,5199.648,1025.404,ribosome,TS_86_3,1 +559.463,4480.427,1114.371,ribosome,TS_86_3,1 +420.204,4238.479,1147.024,ribosome,TS_86_3,1 +92.928,4298.622,1089.994,ribosome,TS_86_3,1 +5020.213,4244.872,791.973,ribosome,TS_86_3,1 +1452.097,5112.11,1022.865,ribosome,TS_86_3,1 +4438.741,4920.707,515.641,ribosome,TS_86_3,1 +4518.449,4937.775,1036.288,ribosome,TS_86_3,1 +4565.882,4348.095,775.964,ribosome,TS_86_3,1 +5572.189,4125.515,785.977,ribosome,TS_86_3,1 +5818.377,4480.871,856.064,ribosome,TS_86_3,1 +5938.142,4347.523,1016.263,ribosome,TS_86_3,1 +5416.05,4499.133,856.064,ribosome,TS_86_3,1 +5931.509,4108.851,866.076,ribosome,TS_86_3,1 +1959.59,4591.854,903.764,ribosome,TS_86_3,1 +1358.159,4893.146,936.164,ribosome,TS_86_3,1 +1285.497,4093.274,632.591,ribosome,TS_86_3,1 +1297.423,4036.911,866.076,ribosome,TS_86_3,1 +1633.782,4376.815,745.927,ribosome,TS_86_3,1 +616.559,5883.402,685.852,ribosome,TS_86_3,1 +513.851,5639.86,836.039,ribosome,TS_86_3,1 +4341.534,4578.56,816.014,ribosome,TS_86_3,1 +3491.458,1964.375,301.542,thyroglobulin,TS_86_3,1 +1059.895,338.01,466.754,thyroglobulin,TS_86_3,1 +5583.535,5055.373,522.266,thyroglobulin,TS_86_3,1 +3753.362,3195.014,497.106,thyroglobulin,TS_86_3,1 +4845.469,5984.182,512.044,thyroglobulin,TS_86_3,1 +3528.827,6079.586,566.737,thyroglobulin,TS_86_3,1 +4527.884,512.538,643.163,thyroglobulin,TS_86_3,1 +5107.291,5147.205,654.024,thyroglobulin,TS_86_3,1 +4062.524,5536.16,724.915,thyroglobulin,TS_86_3,1 +4426.324,5717.765,731.724,thyroglobulin,TS_86_3,1 +381.507,338.051,836.647,thyroglobulin,TS_86_3,1 +4184.942,666.409,809.653,thyroglobulin,TS_86_3,1 +1339.037,3068.829,802.446,thyroglobulin,TS_86_3,1 +2420.207,148.637,863.583,thyroglobulin,TS_86_3,1 +4547.914,2728.749,949.698,thyroglobulin,TS_86_3,1 +1358.067,798.14,896.974,thyroglobulin,TS_86_3,1 +5038.553,1656.659,993.759,thyroglobulin,TS_86_3,1 +3840.899,5011.353,972.844,thyroglobulin,TS_86_3,1 +4325.738,386.706,1027.049,thyroglobulin,TS_86_3,1 +3082.518,2718.95,1080.23,thyroglobulin,TS_86_3,1 +1856.208,5223.976,1135.201,thyroglobulin,TS_86_3,1 +1212.201,2091.175,1244.156,thyroglobulin,TS_86_3,1 +5353.462,5348.654,880.096,thyroglobulin,TS_86_3,1 +2431.176,5760.392,683.137,thyroglobulin,TS_86_3,1 +5853.226,2606.774,184.839,thyroglobulin,TS_86_3,1 +3131.439,1631.365,1285.867,thyroglobulin,TS_86_3,1 +4843.009,1838.761,1023.053,thyroglobulin,TS_86_3,1 +3472.222,325.556,845.556,thyroglobulin,TS_86_3,1 +3170.0,370.0,940.0,thyroglobulin,TS_86_3,1 +2042.034,923.559,1130.339,thyroglobulin,TS_86_3,1 +2640.654,2140.759,734.476,thyroglobulin,TS_86_3,1 +1979.6,2052.4,692.8,thyroglobulin,TS_86_3,1 +3782.069,3580.345,988.621,thyroglobulin,TS_86_3,1 +394.643,2931.786,1176.786,thyroglobulin,TS_86_3,1 +2799.672,5092.022,408.47,thyroglobulin,TS_86_3,1 +4646.594,1023.948,467.896,thyroglobulin,TS_86_3,1 +1756.142,3207.48,651.26,thyroglobulin,TS_86_3,1 +2860.196,2950.392,609.216,thyroglobulin,TS_86_3,1 +1544.041,3483.215,690.914,thyroglobulin,TS_86_3,1 +5901.04,600.347,761.965,thyroglobulin,TS_86_3,1 +4904.507,2199.128,875.641,thyroglobulin,TS_86_3,1 +3333.953,3082.093,905.233,thyroglobulin,TS_86_3,1 +3923.75,5790.0,952.5,thyroglobulin,TS_86_3,1 +3840.601,1873.09,970.687,thyroglobulin,TS_86_3,1 +5164.483,2126.724,1004.655,thyroglobulin,TS_86_3,1 +154.098,4542.461,671.862,virus-like-particle,TS_86_3,1 +5162.937,6250.567,783.617,virus-like-particle,TS_86_3,1 +5478.422,5660.856,1001.878,virus-like-particle,TS_86_3,1 +6024.374,5816.326,630.589,virus-like-particle,TS_86_3,1 +5670.599,3861.008,454.412,virus-like-particle,TS_86_3,1 +6097.154,3365.906,615.748,virus-like-particle,TS_86_3,1 +5623.065,2769.741,432.341,virus-like-particle,TS_86_3,1 +3895.295,4743.936,724.292,virus-like-particle,TS_86_3,1 +3921.092,4158.7,972.267,virus-like-particle,TS_86_3,1 +2720.586,4840.967,1057.297,virus-like-particle,TS_86_3,1 +984.028,881.454,770.454,virus-like-particle,TS_86_3,1 +796.253,1232.922,956.509,virus-like-particle,TS_86_3,1 +1212.736,396.454,913.84,virus-like-particle,TS_86_3,1 +1234.637,1416.69,758.009,virus-like-particle,TS_86_3,1 +713.243,3300.755,729.3,virus-like-particle,TS_86_3,1 +449.787,3348.949,936.699,virus-like-particle,TS_86_3,1 +1024.638,2983.1,649.246,virus-like-particle,TS_86_3,1 +640.315,2221.381,868.266,virus-like-particle,TS_86_3,1 +3443.658,858.443,609.676,virus-like-particle,TS_86_3,1 +2733.056,387.938,935.431,virus-like-particle,TS_86_3,1 +3993.256,185.537,625.067,virus-like-particle,TS_86_3,1 +2891.009,1717.577,755.086,virus-like-particle,TS_86_3,1 +2785.126,2361.186,1008.411,virus-like-particle,TS_86_3,1 +4638.201,1844.865,491.223,virus-like-particle,TS_86_3,1 +4683.103,1546.998,899.626,virus-like-particle,TS_86_3,1 +3563.17,2650.076,660.386,virus-like-particle,TS_86_3,1 +3994.606,2797.533,839.291,virus-like-particle,TS_86_3,1 +3829.926,2129.623,859.549,virus-like-particle,TS_86_3,1 +1905.062,2172.381,978.064,virus-like-particle,TS_86_3,1 +3870.343,4952.714,1261.6,apo-ferritin,TS_86_3,1 +4130.897,5422.292,501.86,apo-ferritin,TS_86_3,1 +2735.0,4668.447,520.291,apo-ferritin,TS_86_3,1 +2649.615,4690.615,600.923,apo-ferritin,TS_86_3,1 +2665.353,4810.641,612.019,apo-ferritin,TS_86_3,1 +2957.522,4630.03,489.701,apo-ferritin,TS_86_3,1 +2920.264,4718.238,659.559,apo-ferritin,TS_86_3,1 +2996.667,4730.0,436.667,apo-ferritin,TS_86_3,1 +2957.199,4845.319,394.858,apo-ferritin,TS_86_3,1 +2858.555,4515.665,523.699,apo-ferritin,TS_86_3,1 +2802.396,4464.345,937.859,apo-ferritin,TS_86_3,1 +3636.645,5702.839,646.419,apo-ferritin,TS_86_3,1 +349.225,4387.254,639.613,apo-ferritin,TS_86_3,1 +1421.215,4837.57,473.209,apo-ferritin,TS_86_3,1 +1516.687,4884.498,530.122,apo-ferritin,TS_86_3,1 +1344.405,5016.071,464.077,apo-ferritin,TS_86_3,1 +1615.455,5224.058,418.604,apo-ferritin,TS_86_3,1 +1529.628,5302.061,442.027,apo-ferritin,TS_86_3,1 +1744.252,5121.395,462.245,apo-ferritin,TS_86_3,1 +1289.961,4419.807,498.842,apo-ferritin,TS_86_3,1 +368.896,5304.785,611.472,apo-ferritin,TS_86_3,1 +349.415,5404.185,663.938,apo-ferritin,TS_86_3,1 +737.13,5657.546,451.944,apo-ferritin,TS_86_3,1 +1152.315,5759.259,463.056,apo-ferritin,TS_86_3,1 +1223.792,5738.917,536.083,apo-ferritin,TS_86_3,1 +1111.487,5848.885,503.234,apo-ferritin,TS_86_3,1 +2024.386,2430.912,928.211,apo-ferritin,TS_86_3,1 +1885.083,2451.761,899.07,apo-ferritin,TS_86_3,1 +1635.436,2052.785,702.483,apo-ferritin,TS_86_3,1 +764.053,3009.015,780.114,apo-ferritin,TS_86_3,1 +805.603,2963.582,867.943,apo-ferritin,TS_86_3,1 +812.748,562.691,574.079,apo-ferritin,TS_86_3,1 +2398.017,1136.609,1115.057,apo-ferritin,TS_86_3,1 +2339.747,1240.886,1132.057,apo-ferritin,TS_86_3,1 +2423.333,1373.903,1099.801,apo-ferritin,TS_86_3,1 +1882.737,1238.321,1279.672,apo-ferritin,TS_86_3,1 +2886.727,662.649,515.325,apo-ferritin,TS_86_3,1 +3243.811,189.939,1204.756,apo-ferritin,TS_86_3,1 +5632.418,4832.451,852.157,apo-ferritin,TS_86_3,1 +5571.34,4880.069,923.814,apo-ferritin,TS_86_3,1 +5786.293,5116.573,858.037,apo-ferritin,TS_86_3,1 +5775.152,4995.212,878.485,apo-ferritin,TS_86_3,1 +4335.3,3517.367,1224.1,apo-ferritin,TS_86_3,1 +4765.574,621.401,780.112,apo-ferritin,TS_86_3,1 +4728.462,597.308,964.615,apo-ferritin,TS_86_3,1 +4784.136,555.39,853.254,apo-ferritin,TS_86_3,1 +3028.839,2861.645,444.29,apo-ferritin,TS_86_3,1 +3098.086,3170.695,1146.425,apo-ferritin,TS_86_3,1 +3192.802,3247.743,1146.425,apo-ferritin,TS_86_3,1 +3123.334,3157.011,1266.574,apo-ferritin,TS_86_3,1 +3197.58,5525.735,1296.612,apo-ferritin,TS_86_3,1 +1912.78,3850.979,1236.537,apo-ferritin,TS_86_3,1 +3760.282,4956.3,1186.475,apo-ferritin,TS_86_3,1 +982.232,4714.162,475.591,apo-ferritin,TS_86_3,1 +996.927,4489.487,475.591,apo-ferritin,TS_86_3,1 +745.902,4018.275,615.765,apo-ferritin,TS_86_3,1 +375.196,4211.727,615.765,apo-ferritin,TS_86_3,1 +258.592,4183.515,645.803,apo-ferritin,TS_86_3,1 +2084.869,5220.131,435.541,apo-ferritin,TS_86_3,1 +718.028,4094.611,565.703,apo-ferritin,TS_86_3,1 +457.252,4137.12,505.628,apo-ferritin,TS_86_3,1 +1361.928,5703.278,435.541,apo-ferritin,TS_86_3,1 +1112.671,5943.359,455.566,apo-ferritin,TS_86_3,1 +69.928,5362.53,555.691,apo-ferritin,TS_86_3,1 +3824.768,5064.901,676.026,beta-amylase,TS_69_2,2 +5166.324,3905.441,924.118,beta-amylase,TS_69_2,2 +5123.125,5389.375,945.625,beta-amylase,TS_69_2,2 +4407.778,6266.944,1228.333,beta-amylase,TS_69_2,2 +350.0,1220.0,1210.0,beta-amylase,TS_69_2,2 +1567.122,2131.079,687.05,beta-amylase,TS_69_2,2 +1466.143,913.857,1258.714,beta-amylase,TS_69_2,2 +2791.339,678.571,1164.821,beta-amylase,TS_69_2,2 +4378.638,762.894,913.191,beta-amylase,TS_69_2,2 +3569.083,427.11,935.688,beta-amylase,TS_69_2,2 +4442.424,2286.136,669.773,beta-amylase,TS_69_2,2 +4263.574,3363.976,1001.928,beta-amylase,TS_69_2,2 +3124.091,2349.758,914.818,beta-galactosidase,TS_69_2,2 +3104.737,116.316,893.158,beta-galactosidase,TS_69_2,2 +1482.15,1976.86,809.963,beta-galactosidase,TS_69_2,2 +4736.857,1633.5,1281.857,beta-galactosidase,TS_69_2,2 +5440.842,2153.316,1065.263,beta-galactosidase,TS_69_2,2 +5829.167,3585.0,1234.167,beta-galactosidase,TS_69_2,2 +5934.359,4038.974,711.538,beta-galactosidase,TS_69_2,2 +5565.034,3118.993,770.604,beta-galactosidase,TS_69_2,2 +1786.416,1087.714,590.104,beta-galactosidase,TS_69_2,2 +5058.789,2709.421,670.105,beta-galactosidase,TS_69_2,2 +1884.097,2071.736,719.583,beta-galactosidase,TS_69_2,2 +5685.556,4460.606,717.172,beta-galactosidase,TS_69_2,2 +4895.506,2074.177,754.272,beta-galactosidase,TS_69_2,2 +3093.723,2528.936,923.617,beta-galactosidase,TS_69_2,2 +5015.0,1340.263,944.649,beta-galactosidase,TS_69_2,2 +5835.63,5542.991,1125.63,beta-galactosidase,TS_69_2,2 +6006.82,5328.856,808.633,ribosome,TS_69_2,2 +5957.767,5571.663,706.068,ribosome,TS_69_2,2 +5618.836,6161.76,693.695,ribosome,TS_69_2,2 +1127.794,187.938,537.361,ribosome,TS_69_2,2 +1767.82,3213.929,1001.62,ribosome,TS_69_2,2 +1916.347,3426.268,1005.688,ribosome,TS_69_2,2 +1171.606,3275.311,1023.63,ribosome,TS_69_2,2 +1475.204,3418.907,1062.102,ribosome,TS_69_2,2 +855.205,1905.764,988.053,ribosome,TS_69_2,2 +1041.483,1639.013,1025.697,ribosome,TS_69_2,2 +588.724,1717.082,887.573,ribosome,TS_69_2,2 +1158.38,2417.84,824.652,ribosome,TS_69_2,2 +3944.471,877.964,875.367,ribosome,TS_69_2,2 +3800.311,509.266,1148.402,ribosome,TS_69_2,2 +5080.617,763.376,705.302,ribosome,TS_69_2,2 +4661.842,201.026,1140.471,ribosome,TS_69_2,2 +4909.496,342.802,905.927,ribosome,TS_69_2,2 +3032.898,1801.65,980.107,ribosome,TS_69_2,2 +3281.41,1786.652,1039.956,ribosome,TS_69_2,2 +3263.628,3307.453,718.927,ribosome,TS_69_2,2 +3204.775,3028.787,664.15,ribosome,TS_69_2,2 +3411.387,3674.855,1015.965,ribosome,TS_69_2,2 +2957.875,3215.688,651.509,ribosome,TS_69_2,2 +2988.816,2839.513,747.089,ribosome,TS_69_2,2 +3759.918,3733.779,875.956,ribosome,TS_69_2,2 +4064.384,3855.712,1073.271,ribosome,TS_69_2,2 +4093.64,2806.907,992.895,ribosome,TS_69_2,2 +4553.26,2854.694,934.778,ribosome,TS_69_2,2 +4833.247,2785.558,1008.777,ribosome,TS_69_2,2 +4982.25,3012.384,1141.249,ribosome,TS_69_2,2 +4746.456,3247.044,989.765,ribosome,TS_69_2,2 +6205.712,3190.031,1166.83,ribosome,TS_69_2,2 +5678.07,3034.789,1201.75,ribosome,TS_69_2,2 +5473.934,5538.159,1008.782,ribosome,TS_69_2,2 +1863.411,1820.198,1074.291,ribosome,TS_69_2,2 +4310.582,2600.476,966.971,ribosome,TS_69_2,2 +3825.977,2568.575,1072.554,ribosome,TS_69_2,2 +1697.664,2484.89,553.359,thyroglobulin,TS_69_2,2 +1083.746,4336.916,660.712,thyroglobulin,TS_69_2,2 +2535.083,3739.817,626.862,thyroglobulin,TS_69_2,2 +719.11,2318.413,647.963,thyroglobulin,TS_69_2,2 +270.918,4900.032,677.824,thyroglobulin,TS_69_2,2 +5371.849,897.815,674.958,thyroglobulin,TS_69_2,2 +5539.777,1677.178,752.772,thyroglobulin,TS_69_2,2 +2021.719,797.706,864.533,thyroglobulin,TS_69_2,2 +2367.94,978.443,867.55,thyroglobulin,TS_69_2,2 +5605.799,1623.846,1155.562,thyroglobulin,TS_69_2,2 +843.974,3537.848,538.51,thyroglobulin,TS_69_2,2 +714.545,3129.091,673.636,thyroglobulin,TS_69_2,2 +1456.572,4357.201,783.836,thyroglobulin,TS_69_2,2 +4435.591,5401.685,874.624,thyroglobulin,TS_69_2,2 +4315.138,4204.558,650.083,thyroglobulin,TS_69_2,2 +3988.304,3924.261,623.783,thyroglobulin,TS_69_2,2 +5049.901,4123.515,1092.673,thyroglobulin,TS_69_2,2 +5901.79,5390.067,1034.295,thyroglobulin,TS_69_2,2 +4435.398,6184.513,718.702,thyroglobulin,TS_69_2,2 +4195.77,2991.516,666.993,thyroglobulin,TS_69_2,2 +5720.0,1579.247,825.146,thyroglobulin,TS_69_2,2 +3050.0,2350.0,810.0,thyroglobulin,TS_69_2,2 +2991.25,1755.417,616.25,thyroglobulin,TS_69_2,2 +3004.667,1165.667,938.333,thyroglobulin,TS_69_2,2 +2035.125,145.5,1198.875,thyroglobulin,TS_69_2,2 +2021.887,1799.321,698.34,thyroglobulin,TS_69_2,2 +763.441,469.624,885.591,thyroglobulin,TS_69_2,2 +6229.592,394.49,1286.327,thyroglobulin,TS_69_2,2 +1369.833,5924.057,828.807,thyroglobulin,TS_69_2,2 +785.587,2630.0,576.201,thyroglobulin,TS_69_2,2 +3362.941,695.882,620.588,thyroglobulin,TS_69_2,2 +3625.942,4047.846,729.154,thyroglobulin,TS_69_2,2 +2618.947,369.774,749.173,thyroglobulin,TS_69_2,2 +5679.653,4176.597,1108.889,thyroglobulin,TS_69_2,2 +235.918,2967.62,799.798,virus-like-particle,TS_69_2,2 +516.962,4021.185,878.484,virus-like-particle,TS_69_2,2 +2832.249,750.827,654.317,virus-like-particle,TS_69_2,2 +2587.613,1360.659,814.115,virus-like-particle,TS_69_2,2 +5063.363,1732.215,951.33,virus-like-particle,TS_69_2,2 +5930.927,978.082,911.592,virus-like-particle,TS_69_2,2 +3603.379,2385.445,924.877,virus-like-particle,TS_69_2,2 +5125.065,5681.883,888.948,virus-like-particle,TS_69_2,2 +5924.779,3598.073,941.761,virus-like-particle,TS_69_2,2 +770.625,1111.161,1088.795,apo-ferritin,TS_69_2,2 +828.291,1201.673,1153.745,apo-ferritin,TS_69_2,2 +668.986,1041.449,1102.246,apo-ferritin,TS_69_2,2 +834.049,592.958,698.099,apo-ferritin,TS_69_2,2 +81.893,2152.929,543.179,apo-ferritin,TS_69_2,2 +5194.716,2277.793,916.254,apo-ferritin,TS_69_2,2 +2720.996,975.578,759.681,apo-ferritin,TS_69_2,2 +409.224,5385.982,819.132,apo-ferritin,TS_69_2,2 +470.992,5471.736,826.116,apo-ferritin,TS_69_2,2 +1340.353,5860.929,518.237,apo-ferritin,TS_69_2,2 +6150.553,4248.018,598.848,apo-ferritin,TS_69_2,2 +4116.256,6243.555,550.142,apo-ferritin,TS_69_2,2 +1695.969,4701.783,576.473,apo-ferritin,TS_69_2,2 +1639.36,4850.438,1063.3,apo-ferritin,TS_69_2,2 +4775.286,5015.714,581.514,apo-ferritin,TS_69_2,2 +4844.739,4397.387,782.334,apo-ferritin,TS_69_2,2 +3192.063,3826.044,607.451,apo-ferritin,TS_69_2,2 +3809.442,4198.247,561.673,apo-ferritin,TS_69_2,2 +3583.403,4366.702,603.141,apo-ferritin,TS_69_2,2 +3476.717,4383.102,650.181,apo-ferritin,TS_69_2,2 +3346.316,4185.936,586.959,apo-ferritin,TS_69_2,2 +3356.477,4340.341,649.659,apo-ferritin,TS_69_2,2 +3288.077,4440.594,634.301,apo-ferritin,TS_69_2,2 +3227.491,4544.364,579.127,apo-ferritin,TS_69_2,2 +3986.778,4435.444,572.259,apo-ferritin,TS_69_2,2 +3838.352,3983.448,633.065,apo-ferritin,TS_69_2,2 +4414.178,3489.079,1138.289,apo-ferritin,TS_69_2,2 +2864.757,2420.825,1037.476,apo-ferritin,TS_69_2,2 +2965.475,2392.793,1081.285,apo-ferritin,TS_69_2,2 +2450.507,2722.905,999.426,apo-ferritin,TS_69_2,2 +3374.729,4537.176,585.728,apo-ferritin,TS_69_2,2 +3463.362,4759.63,585.728,apo-ferritin,TS_69_2,2 +3280.785,4270.741,665.828,apo-ferritin,TS_69_2,2 +3738.087,4294.279,665.828,apo-ferritin,TS_69_2,2 +3811.914,4381.302,616.461,apo-ferritin,TS_69_2,2 +2913.77,5921.278,867.604,beta-amylase,TS_99_9,3 +2806.057,5772.0,1138.4,beta-amylase,TS_99_9,3 +1676.875,5103.125,1180.0,beta-amylase,TS_99_9,3 +579.608,4457.059,1065.686,beta-amylase,TS_99_9,3 +920.0,5645.0,1530.0,beta-amylase,TS_99_9,3 +4033.412,3769.948,1173.543,beta-amylase,TS_99_9,3 +2777.874,3491.181,478.15,beta-amylase,TS_99_9,3 +1885.833,236.667,184.167,beta-amylase,TS_99_9,3 +1999.49,2273.469,310.663,beta-amylase,TS_99_9,3 +2345.73,2017.64,996.966,beta-amylase,TS_99_9,3 +2080.0,1815.0,311.275,beta-amylase,TS_99_9,3 +3379.651,1627.093,693.682,beta-amylase,TS_99_9,3 +2832.0,1580.0,670.0,beta-amylase,TS_99_9,3 +3680.78,2053.756,450.488,beta-amylase,TS_99_9,3 +1990.0,2810.0,730.0,beta-amylase,TS_99_9,3 +3614.741,2640.948,742.026,beta-amylase,TS_99_9,3 +1571.774,1085.222,661.613,beta-amylase,TS_99_9,3 +1440.667,1001.667,790.333,beta-amylase,TS_99_9,3 +2501.623,951.948,806.623,beta-amylase,TS_99_9,3 +730.0,1848.333,921.667,beta-amylase,TS_99_9,3 +4443.627,6262.353,945.882,beta-amylase,TS_99_9,3 +5628.926,4904.558,596.945,beta-galactosidase,TS_99_9,3 +5173.756,5124.78,558.683,beta-galactosidase,TS_99_9,3 +3510.155,195.58,743.106,beta-galactosidase,TS_99_9,3 +5368.571,3285.0,924.286,beta-galactosidase,TS_99_9,3 +5870.914,2555.995,402.796,beta-galactosidase,TS_99_9,3 +5692.178,363.77,822.506,beta-galactosidase,TS_99_9,3 +4330.594,227.707,232.155,beta-galactosidase,TS_99_9,3 +2962.0,1081.667,524.222,beta-galactosidase,TS_99_9,3 +2595.909,762.727,660.0,beta-galactosidase,TS_99_9,3 +2189.81,932.048,518.571,beta-galactosidase,TS_99_9,3 +2145.786,1449.929,283.357,beta-galactosidase,TS_99_9,3 +485.221,2212.794,529.044,beta-galactosidase,TS_99_9,3 +761.031,2527.937,656.812,beta-galactosidase,TS_99_9,3 +1503.711,5532.695,1127.148,beta-galactosidase,TS_99_9,3 +3130.0,3430.0,325.0,beta-galactosidase,TS_99_9,3 +1542.335,4102.275,970.0,beta-galactosidase,TS_99_9,3 +430.436,5762.282,972.349,beta-galactosidase,TS_99_9,3 +6028.531,1814.84,489.704,beta-galactosidase,TS_99_9,3 +5142.516,1120.644,610.598,beta-galactosidase,TS_99_9,3 +2907.5,3320.0,580.0,beta-galactosidase,TS_99_9,3 +3112.493,2771.804,800.133,beta-galactosidase,TS_99_9,3 +2058.911,3929.241,786.634,beta-galactosidase,TS_99_9,3 +807.718,4597.49,1060.788,beta-galactosidase,TS_99_9,3 +5390.233,5613.953,1080.233,beta-galactosidase,TS_99_9,3 +3319.52,5476.3,1014.511,ribosome,TS_99_9,3 +3106.079,5574.209,1186.11,ribosome,TS_99_9,3 +3676.997,5271.281,890.752,ribosome,TS_99_9,3 +3717.18,5455.125,1103.21,ribosome,TS_99_9,3 +3525.117,5614.423,1112.576,ribosome,TS_99_9,3 +3535.323,5008.014,1059.689,ribosome,TS_99_9,3 +2883.189,5382.516,1133.506,ribosome,TS_99_9,3 +2116.336,5189.631,1027.456,ribosome,TS_99_9,3 +2328.909,5130.186,1253.405,ribosome,TS_99_9,3 +1947.981,5391.709,1313.177,ribosome,TS_99_9,3 +2156.805,5609.771,1273.522,ribosome,TS_99_9,3 +2509.295,5370.443,1185.571,ribosome,TS_99_9,3 +2719.011,5096.74,1263.963,ribosome,TS_99_9,3 +3176.477,5049.588,1126.434,ribosome,TS_99_9,3 +4934.962,6071.261,947.376,ribosome,TS_99_9,3 +4999.826,5866.993,1122.244,ribosome,TS_99_9,3 +4927.254,5595.834,1145.519,ribosome,TS_99_9,3 +5615.95,4909.438,989.317,ribosome,TS_99_9,3 +5336.302,4963.521,1045.105,ribosome,TS_99_9,3 +5279.465,4564.465,417.07,ribosome,TS_99_9,3 +4916.58,5065.955,1120.492,ribosome,TS_99_9,3 +411.526,315.634,370.328,ribosome,TS_99_9,3 +430.518,110.647,133.701,ribosome,TS_99_9,3 +702.557,115.463,165.955,ribosome,TS_99_9,3 +1055.336,457.84,237.445,ribosome,TS_99_9,3 +1152.917,482.744,468.791,ribosome,TS_99_9,3 +1424.626,537.985,227.71,ribosome,TS_99_9,3 +852.272,670.775,246.253,ribosome,TS_99_9,3 +1135.8,1200.179,278.969,ribosome,TS_99_9,3 +833.018,1211.915,291.627,ribosome,TS_99_9,3 +521.382,991.567,232.765,ribosome,TS_99_9,3 +724.625,906.686,436.445,ribosome,TS_99_9,3 +971.899,924.657,189.731,ribosome,TS_99_9,3 +309.802,1409.819,659.974,ribosome,TS_99_9,3 +544.924,1197.59,831.877,ribosome,TS_99_9,3 +539.079,1402.459,906.531,ribosome,TS_99_9,3 +1131.172,171.919,184.836,ribosome,TS_99_9,3 +271.44,4207.943,1181.704,ribosome,TS_99_9,3 +642.942,3879.471,1147.333,ribosome,TS_99_9,3 +559.039,3663.243,1180.24,ribosome,TS_99_9,3 +807.849,4177.77,1283.998,ribosome,TS_99_9,3 +283.913,3819.565,1156.087,ribosome,TS_99_9,3 +256.686,5246.703,1066.913,ribosome,TS_99_9,3 +490.506,5275.121,1291.868,ribosome,TS_99_9,3 +882.72,5136.226,1305.376,ribosome,TS_99_9,3 +2003.082,3328.973,1001.08,ribosome,TS_99_9,3 +1873.277,3543.216,678.046,ribosome,TS_99_9,3 +1118.468,3619.591,1267.875,ribosome,TS_99_9,3 +956.16,2418.385,833.722,ribosome,TS_99_9,3 +842.784,2288.479,463.814,ribosome,TS_99_9,3 +1008.88,2721.708,938.224,ribosome,TS_99_9,3 +1534.252,2653.747,894.814,ribosome,TS_99_9,3 +1766.353,2563.641,1128.742,ribosome,TS_99_9,3 +1713.554,2870.877,1082.887,ribosome,TS_99_9,3 +2887.071,2920.025,1138.122,ribosome,TS_99_9,3 +5283.409,1249.497,917.141,ribosome,TS_99_9,3 +5762.274,1749.037,772.82,ribosome,TS_99_9,3 +5903.842,1565.63,977.371,ribosome,TS_99_9,3 +5482.753,1827.704,848.819,ribosome,TS_99_9,3 +4787.941,1452.551,904.813,ribosome,TS_99_9,3 +4561.989,1641.32,954.457,ribosome,TS_99_9,3 +4166.224,1356.023,1062.256,ribosome,TS_99_9,3 +4306.506,2049.676,1092.797,ribosome,TS_99_9,3 +258.14,5412.14,1270.491,ribosome,TS_99_9,3 +3763.052,5198.677,1156.437,ribosome,TS_99_9,3 +638.516,1499.883,279.57,thyroglobulin,TS_99_9,3 +2038.81,214.041,361.807,thyroglobulin,TS_99_9,3 +1504.54,3662.02,566.838,thyroglobulin,TS_99_9,3 +2166.658,2000.169,547.046,thyroglobulin,TS_99_9,3 +656.526,1988.951,597.846,thyroglobulin,TS_99_9,3 +3082.001,2240.364,645.634,thyroglobulin,TS_99_9,3 +2504.842,4565.01,628.011,thyroglobulin,TS_99_9,3 +5788.387,2978.892,692.567,thyroglobulin,TS_99_9,3 +5360.187,4157.772,727.216,thyroglobulin,TS_99_9,3 +2245.603,1789.767,785.667,thyroglobulin,TS_99_9,3 +1398.087,3775.91,1004.103,thyroglobulin,TS_99_9,3 +3817.136,1389.497,1017.085,thyroglobulin,TS_99_9,3 +1229.988,2017.224,1143.411,thyroglobulin,TS_99_9,3 +2169.349,329.172,1145.444,thyroglobulin,TS_99_9,3 +4890.0,4570.0,1210.0,thyroglobulin,TS_99_9,3 +4649.024,4763.902,624.268,thyroglobulin,TS_99_9,3 +5513.22,4635.311,759.492,thyroglobulin,TS_99_9,3 +4347.313,4142.09,1080.597,thyroglobulin,TS_99_9,3 +3747.657,4574.435,1079.079,thyroglobulin,TS_99_9,3 +5409.036,5614.819,878.072,thyroglobulin,TS_99_9,3 +5141.848,2562.391,982.609,thyroglobulin,TS_99_9,3 +4465.707,2667.488,1096.463,thyroglobulin,TS_99_9,3 +3957.731,2284.118,393.025,thyroglobulin,TS_99_9,3 +3382.5,2805.0,1050.0,thyroglobulin,TS_99_9,3 +3261.336,2442.708,1020.181,thyroglobulin,TS_99_9,3 +2357.384,3572.778,1085.972,thyroglobulin,TS_99_9,3 +1649.074,3786.667,624.074,thyroglobulin,TS_99_9,3 +1905.325,3728.049,1032.276,thyroglobulin,TS_99_9,3 +1498.586,4220.741,1323.401,thyroglobulin,TS_99_9,3 +2345.0,5647.5,895.0,thyroglobulin,TS_99_9,3 +2303.101,5998.481,899.873,thyroglobulin,TS_99_9,3 +3525.787,183.38,733.264,thyroglobulin,TS_99_9,3 +2731.186,1063.66,237.062,thyroglobulin,TS_99_9,3 +1054.579,3043.832,715.813,thyroglobulin,TS_99_9,3 +430.0,2690.0,930.0,thyroglobulin,TS_99_9,3 +1503.692,1698.051,288.667,thyroglobulin,TS_99_9,3 +676.591,834.773,70.909,thyroglobulin,TS_99_9,3 +492.857,798.312,686.104,thyroglobulin,TS_99_9,3 +452.931,1866.293,293.448,thyroglobulin,TS_99_9,3 +1604.444,695.079,1036.984,thyroglobulin,TS_99_9,3 +2620.97,1504.97,424.606,thyroglobulin,TS_99_9,3 +2576.97,774.646,642.323,thyroglobulin,TS_99_9,3 +2487.147,4256.221,704.01,thyroglobulin,TS_99_9,3 +2280.316,3380.752,791.189,thyroglobulin,TS_99_9,3 +2790.0,3945.0,790.0,thyroglobulin,TS_99_9,3 +1110.308,5734.615,862.154,thyroglobulin,TS_99_9,3 +3148.098,3234.683,952.634,thyroglobulin,TS_99_9,3 +1541.778,4082.63,989.996,thyroglobulin,TS_99_9,3 +1267.073,4858.78,1068.293,thyroglobulin,TS_99_9,3 +2766.389,185.542,510.129,virus-like-particle,TS_99_9,3 +3084.893,487.084,760.823,virus-like-particle,TS_99_9,3 +3693.973,2932.983,620.271,virus-like-particle,TS_99_9,3 +3010.666,2469.137,884.272,virus-like-particle,TS_99_9,3 +556.664,3167.112,664.183,virus-like-particle,TS_99_9,3 +289.388,4113.399,765.344,virus-like-particle,TS_99_9,3 +981.03,1910.541,780.365,virus-like-particle,TS_99_9,3 +1566.401,4710.311,791.778,virus-like-particle,TS_99_9,3 +2010.056,4752.618,1057.078,virus-like-particle,TS_99_9,3 +2244.068,4310.063,959.548,virus-like-particle,TS_99_9,3 +804.27,5817.135,579.493,virus-like-particle,TS_99_9,3 +4198.228,5534.578,858.169,virus-like-particle,TS_99_9,3 +5698.538,3331.427,806.002,virus-like-particle,TS_99_9,3 +6072.464,4038.0,715.679,apo-ferritin,TS_99_9,3 +5967.452,4228.213,1124.601,apo-ferritin,TS_99_9,3 +5847.622,5066.71,678.893,apo-ferritin,TS_99_9,3 +472.853,5632.618,580.676,apo-ferritin,TS_99_9,3 +564.507,5604.88,678.16,apo-ferritin,TS_99_9,3 +890.414,5615.759,467.69,apo-ferritin,TS_99_9,3 +5613.561,672.73,265.579,apo-ferritin,TS_99_9,3 +458.087,1972.013,765.906,apo-ferritin,TS_99_9,3 +1284.111,1752.411,250.87,apo-ferritin,TS_99_9,3 +1284.136,1713.797,367.525,apo-ferritin,TS_99_9,3 +1233.112,2053.458,251.902,apo-ferritin,TS_99_9,3 +1194.107,1911.285,267.179,apo-ferritin,TS_99_9,3 +5543.161,2879.598,408.42,apo-ferritin,TS_99_9,3 +5485.217,2774.62,411.413,apo-ferritin,TS_99_9,3 +5448.11,3486.46,985.533,apo-ferritin,TS_99_9,3 +3529.611,2919.728,380.389,apo-ferritin,TS_99_9,3 +3438.216,3010.29,397.925,apo-ferritin,TS_99_9,3 +4251.254,845.627,720.143,apo-ferritin,TS_99_9,3 +4592.044,1321.788,322.883,apo-ferritin,TS_99_9,3 +4250.071,1837.26,319.929,apo-ferritin,TS_99_9,3 +4395.13,1892.857,352.013,apo-ferritin,TS_99_9,3 +3670.295,1851.882,328.745,apo-ferritin,TS_99_9,3 +3287.917,1286.28,282.56,apo-ferritin,TS_99_9,3 +2967.331,1646.264,312.781,apo-ferritin,TS_99_9,3 +2987.912,1729.794,397.471,apo-ferritin,TS_99_9,3 +652.0,4574.214,416.571,apo-ferritin,TS_99_9,3 +1032.835,4506.417,1003.228,apo-ferritin,TS_99_9,3 +1100.435,4449.249,1078.617,apo-ferritin,TS_99_9,3 +929.932,4524.384,915.137,apo-ferritin,TS_99_9,3 +1189.237,4882.468,464.122,apo-ferritin,TS_99_9,3 +1649.873,4985.285,473.544,apo-ferritin,TS_99_9,3 +2165.042,4617.759,846.807,apo-ferritin,TS_99_9,3 +2215.952,4694.502,928.61,apo-ferritin,TS_99_9,3 +2276.394,4608.261,802.558,apo-ferritin,TS_99_9,3 +2137.354,3437.938,600.0,apo-ferritin,TS_99_9,3 +1685.81,3815.229,1354.985,apo-ferritin,TS_99_9,3 +1962.647,358.235,376.029,beta-amylase,TS_73_6,4 +2212.609,450.632,641.739,beta-amylase,TS_73_6,4 +2177.5,205.0,907.5,beta-amylase,TS_73_6,4 +943.718,1273.397,132.821,beta-amylase,TS_73_6,4 +2271.379,1911.034,546.293,beta-amylase,TS_73_6,4 +4376.951,2160.122,396.829,beta-amylase,TS_73_6,4 +3761.98,2860.0,1021.287,beta-amylase,TS_73_6,4 +2534.933,3468.622,646.267,beta-amylase,TS_73_6,4 +1948.482,3600.052,965.864,beta-amylase,TS_73_6,4 +1949.783,4223.261,241.304,beta-amylase,TS_73_6,4 +1838.74,5085.118,893.386,beta-amylase,TS_73_6,4 +270.861,5594.658,815.646,beta-amylase,TS_73_6,4 +2285.357,2137.571,596.214,beta-galactosidase,TS_73_6,4 +4008.603,383.309,798.529,beta-galactosidase,TS_73_6,4 +839.231,1602.692,1088.077,beta-galactosidase,TS_73_6,4 +188.815,2167.37,769.741,beta-galactosidase,TS_73_6,4 +742.788,2808.077,246.058,beta-galactosidase,TS_73_6,4 +3552.222,228.148,556.481,beta-galactosidase,TS_73_6,4 +4341.167,3756.5,532.667,beta-galactosidase,TS_73_6,4 +3670.099,2118.144,520.99,beta-galactosidase,TS_73_6,4 +2349.623,474.663,334.96,beta-galactosidase,TS_73_6,4 +3725.132,3520.212,537.937,beta-galactosidase,TS_73_6,4 +2132.535,2661.582,689.108,beta-galactosidase,TS_73_6,4 +388.433,3230.336,768.209,beta-galactosidase,TS_73_6,4 +4555.0,3188.75,868.75,beta-galactosidase,TS_73_6,4 +277.276,3258.78,897.378,beta-galactosidase,TS_73_6,4 +5812.21,954.718,134.827,ribosome,TS_73_6,4 +5598.248,890.665,230.051,ribosome,TS_73_6,4 +5869.463,488.783,285.901,ribosome,TS_73_6,4 +5850.904,575.293,735.879,ribosome,TS_73_6,4 +5940.845,736.687,114.293,ribosome,TS_73_6,4 +5600.385,584.159,361.892,ribosome,TS_73_6,4 +5755.45,2294.051,667.695,ribosome,TS_73_6,4 +6045.567,2430.597,715.623,ribosome,TS_73_6,4 +5923.825,2647.136,880.662,ribosome,TS_73_6,4 +5612.53,2504.846,757.101,ribosome,TS_73_6,4 +2975.042,5637.426,1086.291,ribosome,TS_73_6,4 +2783.415,5731.957,917.101,ribosome,TS_73_6,4 +3067.628,5861.795,977.738,ribosome,TS_73_6,4 +6067.043,3899.28,953.521,ribosome,TS_73_6,4 +4227.966,4408.917,1033.563,ribosome,TS_73_6,4 +4124.483,4142.615,1148.458,ribosome,TS_73_6,4 +3880.999,3552.474,1003.724,ribosome,TS_73_6,4 +4595.975,3413.592,1095.442,ribosome,TS_73_6,4 +2492.225,4304.511,735.979,ribosome,TS_73_6,4 +2141.774,4385.963,846.593,ribosome,TS_73_6,4 +2316.45,4391.133,1066.221,ribosome,TS_73_6,4 +2282.211,4170.642,807.039,ribosome,TS_73_6,4 +2155.564,4599.751,1226.02,ribosome,TS_73_6,4 +1845.342,4530.955,856.591,ribosome,TS_73_6,4 +1698.475,4467.553,1310.344,ribosome,TS_73_6,4 +2963.526,4115.076,947.772,ribosome,TS_73_6,4 +3230.411,3924.05,903.202,ribosome,TS_73_6,4 +3501.536,3769.701,1211.88,ribosome,TS_73_6,4 +4136.679,1185.915,793.199,ribosome,TS_73_6,4 +4244.615,1461.256,917.947,ribosome,TS_73_6,4 +4401.817,1002.721,835.005,ribosome,TS_73_6,4 +3898.451,1614.824,1008.257,ribosome,TS_73_6,4 +3506.524,1792.811,1036.299,ribosome,TS_73_6,4 +3056.445,2174.747,1036.756,ribosome,TS_73_6,4 +2259.0,1224.141,715.852,ribosome,TS_73_6,4 +2311.395,1129.953,423.085,ribosome,TS_73_6,4 +2652.943,1048.475,883.276,ribosome,TS_73_6,4 +2435.147,921.61,1049.207,ribosome,TS_73_6,4 +2677.436,868.552,342.294,ribosome,TS_73_6,4 +2587.987,1253.328,771.87,ribosome,TS_73_6,4 +2214.444,2556.385,1187.927,ribosome,TS_73_6,4 +2213.117,2918.397,1252.563,ribosome,TS_73_6,4 +1535.442,3177.381,1205.003,ribosome,TS_73_6,4 +1010.903,3442.274,1142.175,ribosome,TS_73_6,4 +3126.686,4057.299,1156.437,ribosome,TS_73_6,4 +2182.44,1225.034,1076.338,ribosome,TS_73_6,4 +1841.78,1057.592,258.848,thyroglobulin,TS_73_6,4 +2387.281,2326.345,331.932,thyroglobulin,TS_73_6,4 +3242.6,452.621,489.296,thyroglobulin,TS_73_6,4 +409.479,1796.38,525.534,thyroglobulin,TS_73_6,4 +4582.511,1738.01,569.479,thyroglobulin,TS_73_6,4 +1402.102,2384.13,834.912,thyroglobulin,TS_73_6,4 +728.817,4151.187,967.208,thyroglobulin,TS_73_6,4 +6183.204,996.602,946.99,thyroglobulin,TS_73_6,4 +2025.333,4083.449,1013.484,thyroglobulin,TS_73_6,4 +3103.444,4679.447,1043.352,thyroglobulin,TS_73_6,4 +2073.311,3496.993,1163.75,thyroglobulin,TS_73_6,4 +743.959,238.379,1177.853,thyroglobulin,TS_73_6,4 +524.99,5021.579,1300.106,thyroglobulin,TS_73_6,4 +2870.0,1590.0,215.0,thyroglobulin,TS_73_6,4 +4842.5,412.5,438.333,thyroglobulin,TS_73_6,4 +901.401,5325.541,774.427,thyroglobulin,TS_73_6,4 +1885.771,2380.697,668.159,thyroglobulin,TS_73_6,4 +1875.447,2614.342,921.947,thyroglobulin,TS_73_6,4 +608.235,2829.608,478.693,thyroglobulin,TS_73_6,4 +910.265,3239.47,907.848,thyroglobulin,TS_73_6,4 +550.896,1484.776,625.075,thyroglobulin,TS_73_6,4 +4329.907,3719.346,547.009,thyroglobulin,TS_73_6,4 +4524.646,1289.394,73.081,thyroglobulin,TS_73_6,4 +2349.196,3677.366,735.089,thyroglobulin,TS_73_6,4 +3302.378,1831.524,794.207,thyroglobulin,TS_73_6,4 +448.929,2528.214,811.667,thyroglobulin,TS_73_6,4 +2694.397,4488.369,818.369,thyroglobulin,TS_73_6,4 +3452.853,2717.301,962.577,thyroglobulin,TS_73_6,4 +3023.031,6098.949,542.343,virus-like-particle,TS_73_6,4 +2133.513,5544.092,839.866,virus-like-particle,TS_73_6,4 +5175.693,1206.756,215.022,virus-like-particle,TS_73_6,4 +4849.137,666.384,412.947,virus-like-particle,TS_73_6,4 +6161.707,2651.433,465.402,virus-like-particle,TS_73_6,4 +6042.997,3306.628,713.333,virus-like-particle,TS_73_6,4 +5654.074,2942.728,61.945,virus-like-particle,TS_73_6,4 +6013.231,1871.433,680.048,virus-like-particle,TS_73_6,4 +2836.001,455.2,275.173,virus-like-particle,TS_73_6,4 +2482.763,295.84,690.206,virus-like-particle,TS_73_6,4 +3530.063,771.996,413.662,virus-like-particle,TS_73_6,4 +2687.924,2725.328,632.393,virus-like-particle,TS_73_6,4 +2782.653,1819.135,1019.265,virus-like-particle,TS_73_6,4 +1002.78,2972.855,700.217,virus-like-particle,TS_73_6,4 +679.125,2770.252,855.0,virus-like-particle,TS_73_6,4 +1379.817,2702.33,783.111,virus-like-particle,TS_73_6,4 +985.214,2133.973,598.405,virus-like-particle,TS_73_6,4 +1436.036,1368.571,989.781,virus-like-particle,TS_73_6,4 +1636.918,3798.904,1006.075,virus-like-particle,TS_73_6,4 +3298.877,4560.393,515.641,virus-like-particle,TS_73_6,4 +3093.113,4729.102,645.803,virus-like-particle,TS_73_6,4 +3343.273,4722.883,735.915,virus-like-particle,TS_73_6,4 +268.662,4730.318,916.115,apo-ferritin,TS_73_6,4 +238.946,4853.061,909.898,apo-ferritin,TS_73_6,4 +83.114,5729.56,1219.524,apo-ferritin,TS_73_6,4 +582.143,2769.968,1076.364,apo-ferritin,TS_73_6,4 +510.389,2157.244,362.438,apo-ferritin,TS_73_6,4 +619.176,1991.61,520.449,apo-ferritin,TS_73_6,4 +1245.704,2002.993,743.873,apo-ferritin,TS_73_6,4 +1290.839,1518.942,797.372,apo-ferritin,TS_73_6,4 +1328.307,1547.677,653.425,apo-ferritin,TS_73_6,4 +1352.893,1445.597,725.912,apo-ferritin,TS_73_6,4 +993.954,1705.229,366.405,apo-ferritin,TS_73_6,4 +751.138,1426.892,458.677,apo-ferritin,TS_73_6,4 +2350.062,2564.704,216.978,apo-ferritin,TS_73_6,4 +1210.503,5687.487,1217.588,apo-ferritin,TS_73_6,4 +1134.771,5679.086,898.514,apo-ferritin,TS_73_6,4 +1385.017,5077.747,364.983,apo-ferritin,TS_73_6,4 +1501.315,5047.829,332.783,apo-ferritin,TS_73_6,4 +5980.404,4244.379,478.416,apo-ferritin,TS_73_6,4 +5388.235,4480.0,84.706,apo-ferritin,TS_73_6,4 +5377.5,4507.5,85.833,apo-ferritin,TS_73_6,4 +5395.425,4746.993,39.085,apo-ferritin,TS_73_6,4 +5415.078,4630.078,54.219,apo-ferritin,TS_73_6,4 +5269.381,4808.761,43.717,apo-ferritin,TS_73_6,4 +5704.167,3290.0,85.833,apo-ferritin,TS_73_6,4 +5543.821,3528.374,47.154,apo-ferritin,TS_73_6,4 +5303.962,3755.786,58.176,apo-ferritin,TS_73_6,4 +1643.592,547.51,582.98,apo-ferritin,TS_73_6,4 +1725.534,494.356,644.849,apo-ferritin,TS_73_6,4 +3173.276,4450.517,1039.138,apo-ferritin,TS_73_6,4 +3121.796,4434.709,264.563,apo-ferritin,TS_73_6,4 +3009.867,4384.053,285.76,apo-ferritin,TS_73_6,4 +2947.351,4494.183,306.881,apo-ferritin,TS_73_6,4 +3410.451,4713.045,295.94,apo-ferritin,TS_73_6,4 +2367.675,4872.213,269.412,apo-ferritin,TS_73_6,4 +2273.256,4937.781,298.732,apo-ferritin,TS_73_6,4 +2295.068,4704.11,382.363,apo-ferritin,TS_73_6,4 +2309.494,4590.127,344.557,apo-ferritin,TS_73_6,4 +2345.93,4470.807,329.193,apo-ferritin,TS_73_6,4 +2468.306,4779.707,273.485,apo-ferritin,TS_73_6,4 +2381.72,4060.32,1021.0,apo-ferritin,TS_73_6,4 +2428.114,4271.673,1335.552,apo-ferritin,TS_73_6,4 +4396.744,5082.692,132.436,apo-ferritin,TS_73_6,4 +4620.519,2441.364,105.357,apo-ferritin,TS_73_6,4 +4618.52,2364.605,202.105,apo-ferritin,TS_73_6,4 +4639.096,2192.289,61.386,apo-ferritin,TS_73_6,4 +4635.698,2262.0,151.66,apo-ferritin,TS_73_6,4 +4192.5,2611.806,149.769,apo-ferritin,TS_73_6,4 +4239.923,2540.192,172.222,apo-ferritin,TS_73_6,4 +4286.667,2535.149,605.498,apo-ferritin,TS_73_6,4 +4349.217,2580.569,697.082,apo-ferritin,TS_73_6,4 +4393.977,2435.568,552.273,apo-ferritin,TS_73_6,4 +5024.323,2631.095,824.726,apo-ferritin,TS_73_6,4 +3229.887,3158.038,1059.283,apo-ferritin,TS_73_6,4 +3335.99,3213.02,1097.624,apo-ferritin,TS_73_6,4 +3086.108,1940.784,475.514,apo-ferritin,TS_73_6,4 +2990.879,1990.733,521.832,apo-ferritin,TS_73_6,4 +3155.737,2509.436,149.875,apo-ferritin,TS_73_6,4 +3287.952,2245.186,149.096,apo-ferritin,TS_73_6,4 +3383.0,2800.036,194.679,apo-ferritin,TS_73_6,4 +3615.771,1045.806,224.552,apo-ferritin,TS_73_6,4 +3646.689,1782.23,88.311,apo-ferritin,TS_73_6,4 +3585.105,1694.895,99.247,apo-ferritin,TS_73_6,4 +3807.282,2263.107,561.197,apo-ferritin,TS_73_6,4 +4331.237,779.541,209.293,apo-ferritin,TS_73_6,4 +4393.259,885.599,236.462,apo-ferritin,TS_73_6,4 +3621.489,341.667,73.652,apo-ferritin,TS_73_6,4 +2798.99,1315.505,241.045,apo-ferritin,TS_73_6,4 +2536.024,3417.62,256.355,apo-ferritin,TS_73_6,4 +1334.486,3988.322,802.226,apo-ferritin,TS_73_6,4 +1085.929,5158.59,465.579,apo-ferritin,TS_73_6,4 +1177.73,5141.064,425.529,apo-ferritin,TS_73_6,4 +1178.346,5227.0,345.429,apo-ferritin,TS_73_6,4 +1504.843,5180.322,355.442,apo-ferritin,TS_73_6,4 +5058.617,2754.639,826.027,apo-ferritin,TS_73_6,4 +3167.563,2035.642,455.566,apo-ferritin,TS_73_6,4 +1573.409,390.881,495.616,apo-ferritin,TS_73_6,4 +1634.084,507.488,455.566,apo-ferritin,TS_73_6,4 +1545.135,468.754,595.74,apo-ferritin,TS_73_6,4 +1584.637,454.698,685.852,apo-ferritin,TS_73_6,4 +1779.966,580.292,485.604,apo-ferritin,TS_73_6,4 +1582.199,582.679,355.442,apo-ferritin,TS_73_6,4 +2192.89,171.392,495.616,apo-ferritin,TS_73_6,4 +2076.604,200.819,575.716,apo-ferritin,TS_73_6,4 +2189.662,117.874,595.74,apo-ferritin,TS_73_6,4 +1957.082,145.518,465.579,apo-ferritin,TS_73_6,4 +1944.65,161.916,585.728,apo-ferritin,TS_73_6,4 +3270.772,3235.528,1016.263,apo-ferritin,TS_73_6,4 +3421.573,1462.485,305.38,apo-ferritin,TS_73_6,4 +3221.982,2918.325,265.33,apo-ferritin,TS_73_6,4 +3391.454,2889.264,335.417,apo-ferritin,TS_73_6,4 +3927.906,4070.523,225.28,apo-ferritin,TS_73_6,4 +4263.185,4747.497,225.28,apo-ferritin,TS_73_6,4 +4122.371,4852.662,225.28,apo-ferritin,TS_73_6,4 +4236.513,2479.339,515.641,apo-ferritin,TS_73_6,4 +4309.621,2389.686,485.604,apo-ferritin,TS_73_6,4 +4806.667,980.0,211.111,beta-amylase,TS_6_4,5 +2405.9,2058.159,762.887,beta-amylase,TS_6_4,5 +1489.928,1645.29,797.246,beta-amylase,TS_6_4,5 +4400.989,5358.575,751.563,beta-amylase,TS_6_4,5 +3791.75,4560.917,811.0,beta-amylase,TS_6_4,5 +3495.477,4704.215,901.938,beta-amylase,TS_6_4,5 +3876.774,5060.387,1484.194,beta-amylase,TS_6_4,5 +4501.951,4290.139,1074.704,beta-amylase,TS_6_4,5 +1318.542,5045.417,1136.875,beta-amylase,TS_6_4,5 +804.615,1977.846,489.385,beta-galactosidase,TS_6_4,5 +2134.942,3647.023,657.082,beta-galactosidase,TS_6_4,5 +5361.881,5472.376,1032.772,beta-galactosidase,TS_6_4,5 +4130.0,3716.667,465.0,beta-galactosidase,TS_6_4,5 +3072.5,4610.0,1090.0,beta-galactosidase,TS_6_4,5 +2151.626,4674.959,1189.35,beta-galactosidase,TS_6_4,5 +2367.485,5196.81,587.178,beta-galactosidase,TS_6_4,5 +2680.0,2415.476,982.381,beta-galactosidase,TS_6_4,5 +250.577,1621.923,850.962,beta-galactosidase,TS_6_4,5 +1322.147,2720.516,952.283,beta-galactosidase,TS_6_4,5 +1514.91,1642.699,821.722,beta-galactosidase,TS_6_4,5 +4483.644,4570.968,1089.313,beta-galactosidase,TS_6_4,5 +5274.903,5288.121,619.798,ribosome,TS_6_4,5 +5493.057,5181.127,726.624,ribosome,TS_6_4,5 +5656.951,5168.655,421.572,ribosome,TS_6_4,5 +5559.324,5426.553,556.878,ribosome,TS_6_4,5 +5312.439,4964.156,602.574,ribosome,TS_6_4,5 +5130.837,5121.036,631.702,ribosome,TS_6_4,5 +5453.083,4696.154,504.224,ribosome,TS_6_4,5 +5635.855,4803.903,706.456,ribosome,TS_6_4,5 +5106.838,4835.263,619.225,ribosome,TS_6_4,5 +5734.44,5547.349,1265.455,ribosome,TS_6_4,5 +5571.792,5790.483,1296.079,ribosome,TS_6_4,5 +285.661,4791.067,779.819,ribosome,TS_6_4,5 +229.331,4316.582,690.387,ribosome,TS_6_4,5 +232.042,4432.901,1135.128,ribosome,TS_6_4,5 +348.532,4570.763,717.301,ribosome,TS_6_4,5 +97.34,4661.182,1144.298,ribosome,TS_6_4,5 +143.939,5029.347,732.139,ribosome,TS_6_4,5 +1273.648,4386.703,631.656,ribosome,TS_6_4,5 +1223.64,4186.768,819.895,ribosome,TS_6_4,5 +1236.034,4674.307,668.476,ribosome,TS_6_4,5 +650.107,3282.821,1144.294,ribosome,TS_6_4,5 +830.635,3537.878,963.135,ribosome,TS_6_4,5 +903.085,3391.971,1118.573,ribosome,TS_6_4,5 +630.784,3622.607,1171.482,ribosome,TS_6_4,5 +440.984,3866.177,643.857,ribosome,TS_6_4,5 +583.565,4051.159,828.723,ribosome,TS_6_4,5 +759.417,4295.623,734.899,ribosome,TS_6_4,5 +898.55,4141.046,883.996,ribosome,TS_6_4,5 +868.611,3893.736,743.916,ribosome,TS_6_4,5 +802.156,4625.141,708.375,ribosome,TS_6_4,5 +326.143,3537.823,937.014,ribosome,TS_6_4,5 +1504.444,3275.214,1257.063,ribosome,TS_6_4,5 +2644.576,4431.906,990.296,ribosome,TS_6_4,5 +2444.258,4567.69,1141.142,ribosome,TS_6_4,5 +2468.644,4101.445,1115.899,ribosome,TS_6_4,5 +3078.08,6072.239,1009.897,ribosome,TS_6_4,5 +3015.779,5943.612,1281.03,ribosome,TS_6_4,5 +3702.491,5596.93,1336.535,ribosome,TS_6_4,5 +1997.267,1411.194,1149.58,ribosome,TS_6_4,5 +884.666,2137.612,880.833,ribosome,TS_6_4,5 +663.969,2316.198,1203.572,ribosome,TS_6_4,5 +914.745,2431.255,1268.81,ribosome,TS_6_4,5 +394.061,658.244,750.429,ribosome,TS_6_4,5 +921.916,278.031,597.735,ribosome,TS_6_4,5 +1004.534,86.606,524.534,ribosome,TS_6_4,5 +1284.212,412.471,737.595,ribosome,TS_6_4,5 +1370.425,534.452,920.29,ribosome,TS_6_4,5 +619.064,850.866,358.126,ribosome,TS_6_4,5 +1015.725,640.198,622.387,ribosome,TS_6_4,5 +896.557,967.706,588.803,ribosome,TS_6_4,5 +1017.985,990.96,798.864,ribosome,TS_6_4,5 +898.732,759.425,1040.914,ribosome,TS_6_4,5 +1860.317,365.754,336.113,ribosome,TS_6_4,5 +1689.471,123.33,410.727,ribosome,TS_6_4,5 +886.465,1363.99,1164.549,ribosome,TS_6_4,5 +5007.198,2893.632,998.903,ribosome,TS_6_4,5 +4847.396,2618.105,908.936,ribosome,TS_6_4,5 +5318.939,2139.613,824.39,ribosome,TS_6_4,5 +5186.69,2325.778,977.738,ribosome,TS_6_4,5 +4891.054,1969.255,839.404,ribosome,TS_6_4,5 +4700.372,2111.558,1051.613,ribosome,TS_6_4,5 +4585.681,1839.145,1089.516,ribosome,TS_6_4,5 +4545.69,1538.897,425.389,ribosome,TS_6_4,5 +4576.092,1520.016,1009.884,ribosome,TS_6_4,5 +4654.602,1197.481,806.34,ribosome,TS_6_4,5 +2926.093,2721.038,955.726,ribosome,TS_6_4,5 +2928.412,2194.561,1051.608,ribosome,TS_6_4,5 +2930.904,1857.678,1081.158,ribosome,TS_6_4,5 +2971.267,127.642,534.313,ribosome,TS_6_4,5 +3098.923,2941.184,836.926,ribosome,TS_6_4,5 +3294.228,6090.415,1326.649,ribosome,TS_6_4,5 +5736.406,6034.086,1176.462,ribosome,TS_6_4,5 +503.142,393.241,485.604,ribosome,TS_6_4,5 +5336.84,2766.878,946.176,ribosome,TS_6_4,5 +5251.785,2090.452,490.516,thyroglobulin,TS_6_4,5 +5704.207,2059.378,481.002,thyroglobulin,TS_6_4,5 +5926.681,1787.182,496.498,thyroglobulin,TS_6_4,5 +4967.648,2249.978,522.505,thyroglobulin,TS_6_4,5 +1242.919,1464.644,581.353,thyroglobulin,TS_6_4,5 +3547.616,4770.646,595.253,thyroglobulin,TS_6_4,5 +4756.011,4447.041,581.386,thyroglobulin,TS_6_4,5 +3899.562,5206.682,624.795,thyroglobulin,TS_6_4,5 +4818.823,5953.736,667.288,thyroglobulin,TS_6_4,5 +910.287,2806.175,729.882,thyroglobulin,TS_6_4,5 +4305.698,3168.196,757.859,thyroglobulin,TS_6_4,5 +2447.604,3079.107,791.723,thyroglobulin,TS_6_4,5 +3238.774,4832.551,929.867,thyroglobulin,TS_6_4,5 +316.535,1807.724,1089.125,thyroglobulin,TS_6_4,5 +2022.746,4815.141,1054.014,thyroglobulin,TS_6_4,5 +6060.0,5590.0,150.0,thyroglobulin,TS_6_4,5 +1479.773,2517.727,424.545,thyroglobulin,TS_6_4,5 +1770.114,2095.398,713.92,thyroglobulin,TS_6_4,5 +1732.632,2841.579,1069.043,thyroglobulin,TS_6_4,5 +867.222,1503.333,658.333,thyroglobulin,TS_6_4,5 +1668.75,1150.0,758.75,thyroglobulin,TS_6_4,5 +1990.417,5721.667,858.333,thyroglobulin,TS_6_4,5 +2613.889,3310.0,727.778,thyroglobulin,TS_6_4,5 +6035.196,3635.735,1175.735,thyroglobulin,TS_6_4,5 +4056.946,3386.256,432.266,thyroglobulin,TS_6_4,5 +3753.041,5168.249,1048.157,thyroglobulin,TS_6_4,5 +6000.663,1943.878,948.52,thyroglobulin,TS_6_4,5 +3553.393,2129.821,561.696,thyroglobulin,TS_6_4,5 +4490.0,4330.0,747.5,thyroglobulin,TS_6_4,5 +5785.405,4142.342,1002.77,thyroglobulin,TS_6_4,5 +911.29,5638.402,671.21,virus-like-particle,TS_6_4,5 +122.332,5627.746,1066.367,virus-like-particle,TS_6_4,5 +1452.865,4795.545,1116.066,virus-like-particle,TS_6_4,5 +5580.108,1240.86,692.222,virus-like-particle,TS_6_4,5 +4765.58,3469.964,689.813,virus-like-particle,TS_6_4,5 +5088.704,4120.923,981.513,virus-like-particle,TS_6_4,5 +4268.076,2814.277,815.446,virus-like-particle,TS_6_4,5 +5211.319,5766.513,877.832,virus-like-particle,TS_6_4,5 +4509.57,5139.077,1161.95,virus-like-particle,TS_6_4,5 +6045.947,2340.359,745.927,virus-like-particle,TS_6_4,5 +616.51,2880.471,1294.039,apo-ferritin,TS_6_4,5 +1099.033,1820.423,371.571,apo-ferritin,TS_6_4,5 +1019.831,1859.831,400.424,apo-ferritin,TS_6_4,5 +959.708,1708.149,606.039,apo-ferritin,TS_6_4,5 +1010.329,1758.816,718.52,apo-ferritin,TS_6_4,5 +6116.31,2609.963,528.155,apo-ferritin,TS_6_4,5 +761.116,4901.459,1025.15,apo-ferritin,TS_6_4,5 +969.876,4757.709,1085.418,apo-ferritin,TS_6_4,5 +1648.828,4760.276,997.759,apo-ferritin,TS_6_4,5 +4984.373,5655.125,993.835,apo-ferritin,TS_6_4,5 +4320.826,5556.391,694.862,apo-ferritin,TS_6_4,5 +3683.627,5175.882,813.922,apo-ferritin,TS_6_4,5 +3448.962,5244.962,861.423,apo-ferritin,TS_6_4,5 +4215.917,4821.592,1046.194,apo-ferritin,TS_6_4,5 +2094.233,3235.89,953.497,apo-ferritin,TS_6_4,5 +4793.069,3457.834,1102.852,apo-ferritin,TS_6_4,5 +4804.171,3440.284,1208.91,apo-ferritin,TS_6_4,5 +4550.54,2147.492,218.857,apo-ferritin,TS_6_4,5 +4495.06,2052.976,296.726,apo-ferritin,TS_6_4,5 +4471.074,2131.577,383.456,apo-ferritin,TS_6_4,5 +4126.963,1933.063,190.969,apo-ferritin,TS_6_4,5 +3576.426,2612.49,304.859,apo-ferritin,TS_6_4,5 +3634.286,2431.136,572.418,apo-ferritin,TS_6_4,5 +3534.854,2407.485,425.439,apo-ferritin,TS_6_4,5 +4094.142,2567.337,295.148,apo-ferritin,TS_6_4,5 +4114.734,2446.095,331.538,apo-ferritin,TS_6_4,5 +3894.125,2456.271,258.185,apo-ferritin,TS_6_4,5 +3752.045,2505.17,411.989,apo-ferritin,TS_6_4,5 +3729.36,2566.163,517.093,apo-ferritin,TS_6_4,5 +3836.309,2673.691,474.128,apo-ferritin,TS_6_4,5 +3846.464,2570.493,545.246,apo-ferritin,TS_6_4,5 +3942.417,2660.695,413.609,apo-ferritin,TS_6_4,5 +3965.042,2591.765,574.37,apo-ferritin,TS_6_4,5 +4067.936,2609.929,620.107,apo-ferritin,TS_6_4,5 +4009.57,2947.043,412.634,apo-ferritin,TS_6_4,5 +4095.423,2885.423,469.801,apo-ferritin,TS_6_4,5 +4037.127,2899.179,280.336,apo-ferritin,TS_6_4,5 +4115.619,2817.426,332.054,apo-ferritin,TS_6_4,5 +4036.695,2773.898,490.254,apo-ferritin,TS_6_4,5 +4132.45,2712.15,541.35,apo-ferritin,TS_6_4,5 +3913.495,2803.827,494.974,apo-ferritin,TS_6_4,5 +4211.583,3077.606,520.463,apo-ferritin,TS_6_4,5 +4120.404,3128.081,582.424,apo-ferritin,TS_6_4,5 +4356.198,2834.341,298.892,apo-ferritin,TS_6_4,5 +4282.932,2913.759,318.195,apo-ferritin,TS_6_4,5 +4243.778,2985.682,420.398,apo-ferritin,TS_6_4,5 +4172.383,2955.235,626.173,apo-ferritin,TS_6_4,5 +4054.277,3025.843,547.47,apo-ferritin,TS_6_4,5 +4070.927,2541.192,935.695,apo-ferritin,TS_6_4,5 +3357.295,2074.139,225.328,apo-ferritin,TS_6_4,5 +2153.09,846.979,252.847,apo-ferritin,TS_6_4,5 +3596.958,2593.013,415.516,apo-ferritin,TS_6_4,5 +3486.946,2616.549,415.516,apo-ferritin,TS_6_4,5 +3631.179,2496.627,415.516,apo-ferritin,TS_6_4,5 +776.293,4893.666,1196.487,apo-ferritin,TS_6_4,5 +940.013,4882.972,1046.3,apo-ferritin,TS_6_4,5 +843.54,4841.021,1116.388,apo-ferritin,TS_6_4,5 +2326.455,5821.083,917.063,apo-ferritin,TS_6_4,5 +3850.0,1010.0,310.0,beta-amylase,TS_6_6,6 +4063.606,1651.538,239.038,beta-amylase,TS_6_6,6 +3229.286,1990.0,745.714,beta-amylase,TS_6_6,6 +3367.549,1766.986,676.789,beta-amylase,TS_6_6,6 +3241.333,2196.667,1108.667,beta-amylase,TS_6_6,6 +2695.942,1422.556,380.064,beta-amylase,TS_6_6,6 +2657.277,1564.851,477.787,beta-amylase,TS_6_6,6 +402.895,4085.0,1514.474,beta-amylase,TS_6_6,6 +1837.341,4133.064,835.607,beta-amylase,TS_6_6,6 +1485.96,3785.563,277.881,beta-amylase,TS_6_6,6 +2183.235,3039.118,670.588,beta-amylase,TS_6_6,6 +5169.63,5237.037,709.259,beta-amylase,TS_6_6,6 +4910.0,4665.714,738.857,beta-amylase,TS_6_6,6 +4580.0,5189.756,1065.528,beta-amylase,TS_6_6,6 +5524.088,3251.226,619.877,beta-galactosidase,TS_6_6,6 +5046.434,1586.382,988.516,beta-galactosidase,TS_6_6,6 +4640.0,1865.0,685.0,beta-galactosidase,TS_6_6,6 +1777.5,3650.417,600.0,beta-galactosidase,TS_6_6,6 +1343.732,5939.859,1039.93,beta-galactosidase,TS_6_6,6 +4634.583,4465.417,806.667,beta-galactosidase,TS_6_6,6 +4791.0,4911.6,473.6,beta-galactosidase,TS_6_6,6 +3827.744,5276.792,487.469,beta-galactosidase,TS_6_6,6 +4020.131,3614.629,916.419,beta-galactosidase,TS_6_6,6 +4002.222,5008.421,1137.719,beta-galactosidase,TS_6_6,6 +2440.282,2449.981,1140.433,beta-galactosidase,TS_6_6,6 +773.644,4975.664,1499.213,ribosome,TS_6_6,6 +1029.655,5119.687,1332.989,ribosome,TS_6_6,6 +2106.09,4846.302,1289.305,ribosome,TS_6_6,6 +2843.254,5123.492,1377.645,ribosome,TS_6_6,6 +3676.182,4058.335,1228.49,ribosome,TS_6_6,6 +5697.427,1950.562,972.48,ribosome,TS_6_6,6 +2452.111,3000.991,1108.533,ribosome,TS_6_6,6 +4305.75,4499.335,1214.481,ribosome,TS_6_6,6 +3740.366,4541.466,1101.761,ribosome,TS_6_6,6 +5405.801,2023.684,725.902,ribosome,TS_6_6,6 +5432.587,1978.296,1026.276,ribosome,TS_6_6,6 +5343.945,2419.647,826.027,ribosome,TS_6_6,6 +5182.757,2111.002,826.027,ribosome,TS_6_6,6 +5442.264,2473.504,1116.388,ribosome,TS_6_6,6 +5702.357,2282.812,916.139,ribosome,TS_6_6,6 +5156.231,2251.003,1036.288,ribosome,TS_6_6,6 +4984.874,4435.309,1086.35,ribosome,TS_6_6,6 +4816.361,4198.518,1076.338,ribosome,TS_6_6,6 +4333.457,4506.473,933.065,ribosome,TS_6_6,6 +3821.058,4739.232,976.213,ribosome,TS_6_6,6 +4248.584,4224.388,1176.462,ribosome,TS_6_6,6 +782.628,5388.786,1436.786,ribosome,TS_6_6,6 +585.528,5169.929,1436.786,ribosome,TS_6_6,6 +4709.027,5996.216,316.811,thyroglobulin,TS_6_6,6 +2822.785,2974.861,443.679,thyroglobulin,TS_6_6,6 +2974.63,5786.818,491.11,thyroglobulin,TS_6_6,6 +4622.227,4238.002,506.417,thyroglobulin,TS_6_6,6 +1743.59,605.843,529.632,thyroglobulin,TS_6_6,6 +4492.1,2145.4,584.3,thyroglobulin,TS_6_6,6 +2408.626,2515.879,632.912,thyroglobulin,TS_6_6,6 +6004.797,1528.455,670.081,thyroglobulin,TS_6_6,6 +6065.226,782.511,706.026,thyroglobulin,TS_6_6,6 +3444.054,1444.976,785.389,thyroglobulin,TS_6_6,6 +4327.106,2286.262,879.085,thyroglobulin,TS_6_6,6 +2092.183,1968.936,946.212,thyroglobulin,TS_6_6,6 +2141.26,4493.054,956.843,thyroglobulin,TS_6_6,6 +5348.617,5636.545,943.527,thyroglobulin,TS_6_6,6 +2505.834,1629.141,1118.847,thyroglobulin,TS_6_6,6 +760.567,4157.394,1146.652,thyroglobulin,TS_6_6,6 +228.897,2923.352,1286.104,thyroglobulin,TS_6_6,6 +524.918,2713.886,1291.522,thyroglobulin,TS_6_6,6 +3302.056,834.556,1336.967,thyroglobulin,TS_6_6,6 +586.238,3201.857,1507.786,thyroglobulin,TS_6_6,6 +3750.577,1852.692,950.385,thyroglobulin,TS_6_6,6 +3766.612,1578.415,588.798,thyroglobulin,TS_6_6,6 +5407.907,1216.163,407.558,thyroglobulin,TS_6_6,6 +3113.699,4522.603,960.342,thyroglobulin,TS_6_6,6 +3466.387,4875.21,408.571,thyroglobulin,TS_6_6,6 +1856.175,4545.519,1185.519,thyroglobulin,TS_6_6,6 +922.5,5110.0,1055.0,thyroglobulin,TS_6_6,6 +1184.054,5135.135,337.568,thyroglobulin,TS_6_6,6 +2511.073,1898.707,818.244,thyroglobulin,TS_6_6,6 +4817.354,4860.628,979.955,thyroglobulin,TS_6_6,6 +4461.316,3580.702,1067.544,thyroglobulin,TS_6_6,6 +2947.812,1297.5,1109.688,thyroglobulin,TS_6_6,6 +4719.918,1853.143,1229.265,thyroglobulin,TS_6_6,6 +1543.079,3649.778,1353.424,thyroglobulin,TS_6_6,6 +970.18,3473.559,1473.604,thyroglobulin,TS_6_6,6 +4113.927,1540.028,957.754,virus-like-particle,TS_6_6,6 +4205.011,514.238,377.245,virus-like-particle,TS_6_6,6 +3546.662,996.468,1199.592,virus-like-particle,TS_6_6,6 +238.569,3475.323,722.892,virus-like-particle,TS_6_6,6 +404.421,2707.26,970.408,virus-like-particle,TS_6_6,6 +5686.068,5829.641,364.486,virus-like-particle,TS_6_6,6 +5086.352,6157.563,696.444,virus-like-particle,TS_6_6,6 +4657.586,5875.914,627.971,virus-like-particle,TS_6_6,6 +5827.109,4997.341,854.623,virus-like-particle,TS_6_6,6 +5157.398,4441.079,683.717,virus-like-particle,TS_6_6,6 +5278.371,3038.921,406.17,virus-like-particle,TS_6_6,6 +3005.1,4116.745,533.11,virus-like-particle,TS_6_6,6 +2829.054,4175.449,828.66,virus-like-particle,TS_6_6,6 +3248.07,4513.06,582.863,virus-like-particle,TS_6_6,6 +2609.876,4569.876,1169.759,virus-like-particle,TS_6_6,6 +2213.287,4135.017,1286.851,virus-like-particle,TS_6_6,6 +3303.905,5697.825,789.744,virus-like-particle,TS_6_6,6 +1008.748,5949.213,1077.303,virus-like-particle,TS_6_6,6 +5749.052,3911.392,275.342,virus-like-particle,TS_6_6,6 +1916.83,3311.797,754.673,apo-ferritin,TS_6_6,6 +1996.861,3231.277,803.577,apo-ferritin,TS_6_6,6 +2206.512,2975.302,1179.674,apo-ferritin,TS_6_6,6 +285.292,1379.331,417.577,apo-ferritin,TS_6_6,6 +753.781,2633.219,973.094,apo-ferritin,TS_6_6,6 +726.176,2559.49,1473.314,apo-ferritin,TS_6_6,6 +747.829,2630.698,1549.302,apo-ferritin,TS_6_6,6 +1123.151,2698.296,725.852,apo-ferritin,TS_6_6,6 +1234.692,2994.751,920.469,apo-ferritin,TS_6_6,6 +332.707,3382.279,1241.909,apo-ferritin,TS_6_6,6 +766.772,3170.667,1543.544,apo-ferritin,TS_6_6,6 +1569.207,2394.552,1299.69,apo-ferritin,TS_6_6,6 +818.107,3625.562,671.657,apo-ferritin,TS_6_6,6 +1146.957,1508.462,679.398,apo-ferritin,TS_6_6,6 +4586.585,476.585,763.78,apo-ferritin,TS_6_6,6 +4528.663,551.489,773.374,apo-ferritin,TS_6_6,6 +4905.12,1090.843,266.175,apo-ferritin,TS_6_6,6 +5344.971,446.0,60.429,apo-ferritin,TS_6_6,6 +5250.109,3132.92,842.628,apo-ferritin,TS_6_6,6 +5066.127,1954.349,293.048,apo-ferritin,TS_6_6,6 +5671.396,2753.784,797.838,apo-ferritin,TS_6_6,6 +5794.353,2457.302,441.691,apo-ferritin,TS_6_6,6 +3355.743,2365.446,1138.614,apo-ferritin,TS_6_6,6 +3101.77,1278.918,264.197,apo-ferritin,TS_6_6,6 +2705.768,1889.625,1354.642,apo-ferritin,TS_6_6,6 +3109.837,1800.782,954.072,apo-ferritin,TS_6_6,6 +4212.422,2013.806,305.882,apo-ferritin,TS_6_6,6 +4057.61,5767.868,576.581,apo-ferritin,TS_6_6,6 +3153.077,4790.538,460.577,apo-ferritin,TS_6_6,6 +2977.845,4634.696,444.254,apo-ferritin,TS_6_6,6 +3332.697,5251.364,720.576,apo-ferritin,TS_6_6,6 +3442.903,5827.204,1080.502,apo-ferritin,TS_6_6,6 +318.185,5000.579,551.158,apo-ferritin,TS_6_6,6 +1776.173,4887.716,1019.877,apo-ferritin,TS_6_6,6 +3578.705,3480.026,1206.5,apo-ferritin,TS_6_6,6 +5680.786,2759.578,936.164,apo-ferritin,TS_6_6,6 +3432.444,1208.43,826.027,apo-ferritin,TS_6_6,6 +3074.149,1859.506,846.052,apo-ferritin,TS_6_6,6 +1949.78,3068.74,1156.437,apo-ferritin,TS_6_6,6 +3687.965,1751.484,1296.612,apo-ferritin,TS_6_6,6 +3684.71,1654.009,1306.624,apo-ferritin,TS_6_6,6 diff --git a/competitions/kaggle/Cryo-ET/1st_place_solution/utils.py b/competitions/kaggle/Cryo-ET/1st_place_solution/utils.py new file mode 100644 index 000000000..2f86c48e1 --- /dev/null +++ b/competitions/kaggle/Cryo-ET/1st_place_solution/utils.py @@ -0,0 +1,573 @@ +# Copyright (c) MONAI Consortium +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import random +import os +import numpy as np +import pandas as pd +import torch +from torch.utils.data import Sampler, RandomSampler, SequentialSampler, DataLoader, WeightedRandomSampler +from torch import nn, optim +from torch.optim.lr_scheduler import LambdaLR +from torch.optim import lr_scheduler +import importlib +import math + +import logging +import pickle + + +def get_linear_schedule_with_warmup(optimizer, num_warmup_steps, num_training_steps, last_epoch=-1): + """ + from https://github.com/huggingface/transformers/blob/main/src/transformers/optimization.py + Create a schedule with a learning rate that decreases linearly from the initial lr set in the optimizer to 0, after + a warmup period during which it increases linearly from 0 to the initial lr set in the optimizer. + Args: + optimizer ([`~torch.optim.Optimizer`]): + The optimizer for which to schedule the learning rate. + num_warmup_steps (`int`): + The number of steps for the warmup phase. + num_training_steps (`int`): + The total number of training steps. + last_epoch (`int`, *optional*, defaults to -1): + The index of the last epoch when resuming training. + Return: + `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule. + """ + + def lr_lambda(current_step: int): + if current_step < num_warmup_steps: + return float(current_step) / float(max(1, num_warmup_steps)) + return max(0.0, float(num_training_steps - current_step) / float(max(1, num_training_steps - num_warmup_steps))) + + return LambdaLR(optimizer, lr_lambda, last_epoch) + + +def get_cosine_schedule_with_warmup(optimizer, num_warmup_steps, num_training_steps, num_cycles=0.5, last_epoch=-1): + """ + from https://github.com/huggingface/transformers/blob/main/src/transformers/optimization.py + Create a schedule with a learning rate that decreases following the values of the cosine function between the + initial lr set in the optimizer to 0, after a warmup period during which it increases linearly between 0 and the + initial lr set in the optimizer. + Args: + optimizer ([`~torch.optim.Optimizer`]): + The optimizer for which to schedule the learning rate. + num_warmup_steps (`int`): + The number of steps for the warmup phase. + num_training_steps (`int`): + The total number of training steps. + num_cycles (`float`, *optional*, defaults to 0.5): + The number of waves in the cosine schedule (the defaults is to just decrease from the max value to 0 + following a half-cosine). + last_epoch (`int`, *optional*, defaults to -1): + The index of the last epoch when resuming training. + Return: + `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule. + """ + + def lr_lambda(current_step): + if current_step < num_warmup_steps: + return float(current_step) / float(max(1, num_warmup_steps)) + progress = float(current_step - num_warmup_steps) / float(max(1, num_training_steps - num_warmup_steps)) + return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress))) + + return LambdaLR(optimizer, lr_lambda, last_epoch) + + +def calc_grad_norm(parameters, norm_type=2.0): + + if isinstance(parameters, torch.Tensor): + parameters = [parameters] + parameters = [p for p in parameters if p.grad is not None] + norm_type = float(norm_type) + if len(parameters) == 0: + return torch.tensor(0.0) + device = parameters[0].grad.device + total_norm = torch.norm( + torch.stack([torch.norm(p.grad.detach(), norm_type).to(device) for p in parameters]), norm_type + ) + if torch.logical_or(total_norm.isnan(), total_norm.isinf()): + total_norm = None + + return total_norm + + +def calc_weight_norm(parameters, norm_type=2.0): + + # l2_loss = 0 + # for param in parameters : + # l2_loss += 0.5 * torch.sum(param ** 2) + # return l2_loss + + if isinstance(parameters, torch.Tensor): + parameters = [parameters] + parameters = [p for p in parameters if p.grad is not None] + norm_type = float(norm_type) + if len(parameters) == 0: + return torch.tensor(0.0) + device = parameters[0].grad.device + + total_norm = torch.stack([torch.norm(p.detach(), norm_type).to(device) for p in parameters]).mean() + if torch.logical_or(total_norm.isnan(), total_norm.isinf()): + total_norm = None + + return total_norm + + +class OrderedDistributedSampler(Sampler): + def __init__(self, dataset, num_replicas=None, rank=None): + if num_replicas is None: + if not dist.is_available(): + raise RuntimeError("Requires distributed package to be available") + num_replicas = dist.get_world_size() + if rank is None: + if not dist.is_available(): + raise RuntimeError("Requires distributed package to be available") + rank = dist.get_rank() + self.dataset = dataset + self.num_replicas = num_replicas + self.rank = rank + self.num_samples = int(math.ceil(len(self.dataset) * 1.0 / self.num_replicas)) + self.total_size = self.num_samples * self.num_replicas + + print("TOTAL SIZE", self.total_size) + + def __iter__(self): + indices = list(range(len(self.dataset))) + + # add extra samples to make it evenly divisible + indices += indices[: (self.total_size - len(indices))] + assert len(indices) == self.total_size + + # subsample + indices = indices[self.rank * self.num_samples : self.rank * self.num_samples + self.num_samples] + print( + "SAMPLES", + self.rank * self.num_samples, + self.rank * self.num_samples + self.num_samples, + ) + assert len(indices) == self.num_samples + + return iter(indices) + + def __len__(self): + return self.num_samples + + +def sync_across_gpus(t, world_size): + torch.distributed.barrier() + gather_t_tensor = [torch.ones_like(t) for _ in range(world_size)] + torch.distributed.all_gather(gather_t_tensor, t) + return torch.cat(gather_t_tensor) + + +def set_seed(seed=1234): + random.seed(seed) + os.environ["PYTHONHASHSEED"] = str(seed) + np.random.seed(seed) + torch.manual_seed(seed) + torch.cuda.manual_seed(seed) + torch.backends.cudnn.deterministic = False + torch.backends.cudnn.benchmark = True + + +def worker_init_fn(worker_id): + np.random.seed(np.random.get_state()[1][0] + worker_id) + + +def get_model(cfg, ds): + Net = importlib.import_module(cfg.model).Net + net = Net(cfg) + if cfg.pretrained_weights is not None: + if type(cfg.pretrained_weights) == list: + cfg.pretrained_weights = cfg.pretrained_weights[cfg.fold] + print(f"{cfg.local_rank}: loading weights from", cfg.pretrained_weights) + state_dict = torch.load(cfg.pretrained_weights, map_location="cpu") + if "model" in state_dict.keys(): + state_dict = state_dict["model"] + state_dict = {key.replace("module.", ""): val for key, val in state_dict.items()} + if cfg.pop_weights is not None: + print(f"popping {cfg.pop_weights}") + to_pop = [] + for key in state_dict: + for item in cfg.pop_weights: + if item in key: + to_pop += [key] + for key in to_pop: + print(f"popping {key}") + state_dict.pop(key) + + net.load_state_dict(state_dict, strict=cfg.pretrained_weights_strict) + print(f"{cfg.local_rank}: weights loaded from", cfg.pretrained_weights) + + return net + + +def create_checkpoint(cfg, model, optimizer, epoch, scheduler=None, scaler=None): + + state_dict = model.state_dict() + if cfg.save_weights_only: + checkpoint = {"model": state_dict} + return checkpoint + + checkpoint = { + "model": state_dict, + "optimizer": optimizer.state_dict(), + "epoch": epoch, + } + + if scheduler is not None: + checkpoint["scheduler"] = scheduler.state_dict() + + if scaler is not None: + checkpoint["scaler"] = scaler.state_dict() + return checkpoint + + +def load_checkpoint(cfg, model, optimizer, scheduler=None, scaler=None): + + print(f"loading ckpt {cfg.resume_from}") + checkpoint = torch.load(cfg.resume_from, map_location="cpu") + model.load_state_dict(checkpoint["model"]) + optimizer.load_state_dict(checkpoint["optimizer"]) + scheduler_dict = checkpoint["scheduler"] + if scaler is not None: + scaler.load_state_dict(checkpoint["scaler"]) + + epoch = checkpoint["epoch"] + return model, optimizer, scheduler_dict, scaler, epoch + + +def get_dataset(df, cfg, mode="train"): + + # modes train, val, index + print(f"Loading {mode} dataset") + + if mode == "train": + dataset = get_train_dataset(df, cfg) + # elif mode == 'train_val': + # dataset = get_val_dataset(df, cfg) + elif mode == "val": + dataset = get_val_dataset(df, cfg) + elif mode == "test": + dataset = get_test_dataset(df, cfg) + else: + pass + return dataset + + +def get_dataloader(ds, cfg, mode="train"): + + if mode == "train": + dl = get_train_dataloader(ds, cfg) + elif mode == "val": + dl = get_val_dataloader(ds, cfg) + elif mode == "test": + dl = get_test_dataloader(ds, cfg) + return dl + + +def get_train_dataset(train_df, cfg): + + train_dataset = cfg.CustomDataset(train_df, cfg, aug=cfg.train_aug, mode="train") + if cfg.data_sample > 0: + train_dataset = torch.utils.data.Subset(train_dataset, np.arange(cfg.data_sample)) + return train_dataset + + +def get_train_dataloader(train_ds, cfg): + + if cfg.distributed: + sampler = torch.utils.data.distributed.DistributedSampler( + train_ds, num_replicas=cfg.world_size, rank=cfg.local_rank, shuffle=True, seed=cfg.seed + ) + else: + try: + if cfg.random_sampler_frac > 0: + + num_samples = int(len(train_ds) * cfg.random_sampler_frac) + sample_weights = train_ds.sample_weights + sampler = WeightedRandomSampler(sample_weights, num_samples=num_samples) + else: + sampler = None + except: + sampler = None + + if cfg.use_custom_batch_sampler: + sampler = RandomSampler(train_ds) + bsampler = CustomBatchSampler(sampler, batch_size=cfg.batch_size, drop_last=cfg.drop_last) + train_dataloader = DataLoader( + train_ds, + batch_sampler=bsampler, + # shuffle=(sampler is None), + # batch_size=cfg.batch_size, + num_workers=cfg.num_workers, + pin_memory=cfg.pin_memory, + collate_fn=cfg.tr_collate_fn, + # drop_last=cfg.drop_last, + worker_init_fn=worker_init_fn, + ) + else: + + train_dataloader = DataLoader( + train_ds, + sampler=sampler, + shuffle=(sampler is None), + batch_size=cfg.batch_size, + num_workers=cfg.num_workers, + pin_memory=cfg.pin_memory, + collate_fn=cfg.tr_collate_fn, + drop_last=cfg.drop_last, + worker_init_fn=worker_init_fn, + ) + print(f"train: dataset {len(train_ds)}, dataloader {len(train_dataloader)}") + return train_dataloader + + +def get_val_dataset(val_df, cfg, allowed_targets=None): + val_dataset = cfg.CustomDataset(val_df, cfg, aug=cfg.val_aug, mode="val") + return val_dataset + + +def get_val_dataloader(val_ds, cfg): + + if cfg.distributed and cfg.eval_ddp: + sampler = OrderedDistributedSampler(val_ds, num_replicas=cfg.world_size, rank=cfg.local_rank) + else: + sampler = SequentialSampler(val_ds) + + if cfg.batch_size_val is not None: + batch_size = cfg.batch_size_val + else: + batch_size = cfg.batch_size + val_dataloader = DataLoader( + val_ds, + sampler=sampler, + batch_size=batch_size, + num_workers=cfg.num_workers, + pin_memory=cfg.pin_memory, + collate_fn=cfg.val_collate_fn, + worker_init_fn=worker_init_fn, + ) + print(f"valid: dataset {len(val_ds)}, dataloader {len(val_dataloader)}") + return val_dataloader + + +def get_test_dataset(test_df, cfg): + test_dataset = cfg.CustomDataset(test_df, cfg, aug=cfg.val_aug, mode="test") + return test_dataset + + +def get_test_dataloader(test_ds, cfg): + + if cfg.distributed and cfg.eval_ddp: + sampler = OrderedDistributedSampler(test_ds, num_replicas=cfg.world_size, rank=cfg.local_rank) + else: + sampler = SequentialSampler(test_ds) + + if cfg.batch_size_val is not None: + batch_size = cfg.batch_size_val + else: + batch_size = cfg.batch_size + test_dataloader = DataLoader( + test_ds, + sampler=sampler, + batch_size=batch_size, + num_workers=cfg.num_workers, + pin_memory=cfg.pin_memory, + collate_fn=cfg.val_collate_fn, + worker_init_fn=worker_init_fn, + ) + print(f"test: dataset {len(test_ds)}, dataloader {len(test_dataloader)}") + return test_dataloader + + +def get_optimizer(model, cfg): + + params = model.parameters() + + if cfg.optimizer == "Adam": + optimizer = optim.Adam(params, lr=cfg.lr, weight_decay=cfg.weight_decay) + + elif cfg.optimizer == "AdamW_plus": + paras = list(model.named_parameters()) + no_decay = ["bias", "LayerNorm.bias"] + params = [ + { + "params": [param for name, param in paras if (not any(nd in name for nd in no_decay))], + "lr": cfg.lr, + "weight_decay": cfg.weight_decay, + }, + { + "params": [param for name, param in paras if (any(nd in name for nd in no_decay))], + "lr": cfg.lr, + "weight_decay": 0.0, + }, + ] + optimizer = optim.AdamW(params, lr=cfg.lr) + + elif cfg.optimizer == "AdamW": + optimizer = optim.AdamW(params, lr=cfg.lr, weight_decay=cfg.weight_decay) + + elif cfg.optimizer == "SGD": + optimizer = optim.SGD( + params, + lr=cfg.lr, + momentum=cfg.sgd_momentum, + nesterov=cfg.sgd_nesterov, + weight_decay=cfg.weight_decay, + ) + + return optimizer + + +def get_scheduler(cfg, optimizer, total_steps): + + if cfg.schedule == "steplr": + scheduler = optim.lr_scheduler.StepLR( + optimizer, + step_size=cfg.epochs_step * (total_steps // cfg.batch_size) // cfg.world_size, + gamma=0.5, + ) + elif cfg.schedule == "cosine": + scheduler = get_cosine_schedule_with_warmup( + optimizer, + num_warmup_steps=cfg.warmup * (total_steps // cfg.batch_size) // cfg.world_size, + num_training_steps=cfg.epochs * (total_steps // cfg.batch_size) // cfg.world_size, + num_cycles=cfg.num_cycles, + ) + elif cfg.schedule == "linear": + scheduler = get_linear_schedule_with_warmup( + optimizer, + num_warmup_steps=0, + num_training_steps=cfg.epochs * (total_steps // cfg.batch_size) // cfg.world_size, + ) + + elif cfg.schedule == "CosineAnnealingLR": + T_max = int(np.ceil(0.5 * total_steps)) + scheduler = lr_scheduler.CosineAnnealingLR(optimizer, T_max=T_max, eta_min=1e-8) + + # print("num_steps", (total_steps // cfg.batch_size) // cfg.world_size) + + else: + scheduler = None + + return scheduler + + +def read_df(fn): + + if "parquet" in fn: + df = pd.read_parquet(fn, engine="fastparquet") + else: + df = pd.read_csv(fn) + return df + + +def get_data(cfg): + + # setup dataset + if type(cfg.train_df) == list: + cfg.train_df = cfg.train_df[cfg.fold] + print(f"reading {cfg.train_df}") + df = read_df(cfg.train_df) + + if cfg.test: + test_df = read_df(cfg.test_df) + else: + test_df = None + + if cfg.val_df: + if type(cfg.val_df) == list: + cfg.val_df = cfg.val_df[cfg.fold] + val_df = read_df(cfg.val_df) + if cfg.fold > -1: + if "fold" in val_df.columns: + val_df = val_df[val_df["fold"] == cfg.fold] + train_df = df[df["fold"] != cfg.fold] + else: + train_df = df + else: + train_df = df + else: + if cfg.fold == -1: + val_df = df[df["fold"] == 0] + else: + val_df = df[df["fold"] == cfg.fold] + + train_df = df[df["fold"] != cfg.fold] + + return train_df, val_df, test_df + + +def upload_s3(cfg): + from boto3.session import Session + import boto3 + + BUCKET_NAME = cfg.s3_bucket_name + ACCESS_KEY = cfg.s3_access_key + SECRET_KEY = cfg.s3_secret_key + session = Session(aws_access_key_id=ACCESS_KEY, aws_secret_access_key=SECRET_KEY) + s3 = session.resource("s3") + + s3.Bucket(BUCKET_NAME).upload_file( + f"{cfg.output_dir}/fold{cfg.fold}/val_data_seed{cfg.seed}.pth", + f"output/{cfg.name}/fold{cfg.fold}/val_data_seed{cfg.seed}.pth", + ) + s3.Bucket(BUCKET_NAME).upload_file( + f"{cfg.output_dir}/fold{cfg.fold}/test_data_seed{cfg.seed}.pth", + f"output/{cfg.name}/fold{cfg.fold}/test_data_seed{cfg.seed}.pth", + ) + s3.Bucket(BUCKET_NAME).upload_file( + f"{cfg.output_dir}/fold{cfg.fold}/submission_seed{cfg.seed}.csv", + f"output/{cfg.name}/fold{cfg.fold}/submission_seed{cfg.seed}.csv", + ) + + +def flatten(t): + return [item for sublist in t for item in sublist] + + +def set_pandas_display(): + pd.set_option("display.max_columns", None) + pd.set_option("display.max_rows", 10000) + pd.set_option("display.width", 10000) + pd.set_option("display.float_format", lambda x: "%.3f" % x) + + +def dumpobj(file, obj): + with open(file, "wb") as handle: + pickle.dump(obj, handle, protocol=pickle.HIGHEST_PROTOCOL) + + +def loadobj(file): + with open(file, "rb") as handle: + return pickle.load(handle) + + +def get_level(level_str): + """get level""" + l_names = {logging.getLevelName(lvl).lower(): lvl for lvl in [10, 20, 30, 40, 50]} # noqa + return l_names.get(level_str.lower(), logging.INFO) + + +def get_logger(name, level_str): + """get logger""" + logger = logging.getLogger(name) + logger.setLevel(get_level(level_str)) + handler = logging.StreamHandler() + handler.setLevel(level_str) + handler.setFormatter( + logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s") + ) # pylint: disable=C0301 # noqa + logger.addHandler(handler) + + return logger diff --git a/figures/partly_Unet.png b/figures/partly_Unet.png new file mode 100644 index 000000000..7f4f72cdb Binary files /dev/null and b/figures/partly_Unet.png differ