Fish Morphology Code

Data ingestion, processing, and analysis code for Cell states beyond transcriptomics: integrating structural organization and gene expression in hiPSC-derived cardiomyocytes

---

Installation

The recommended installation procedure is to create a conda environment and then pip install into that environment.

Normal users

Clone this repository, then

cd fish_morphology_code
conda create --name cardio_fish_data python=3.7
conda activate cardio_fish_data
pip install -e .

Developers

After following the "Normal users" installation instructions,

pip install -e .[all]
pre-commit install

Downloading the data

See the open data release (produced by this code) for instructions to download the data:

Notes on using the code

Running the auto-contrasting code

To run the image normalization code, from the main repo dir:

contrast_and_segment quilt_data/metadata.csv quilt_data/supporting_files/channel_defs.json --out_dir=output_data

Running cellprofiler to calculate single cell shape and texture features

Before running cellprofiler, download auto-contrasted images from quilt (ex. download_2D_contrasted --test=True) into current working directory.

Create an image set list in format accepted by cellprofiler's LoadData module.

make_cellprofiler_image_set \
    --image_csv ./quilt_data_contrasted_test/metadata.csv \
    --defaults_json fish_morphology_code/cellprofiler/cellprofiler_image_set_defaults.json \
    --path_key rescaled_2D_fov_tiff_path \
    --local_path ./quilt_data_contrasted_test \
    --out_loc ./test_image_set_list.csv

Run cellprofiler pipeline in this repository using test image set list as input:

#!/bin/bash
mkdir cp_out

cellprofiler \
    -p fish_morphology_code/cellprofiler/cp_3i_image_processing.cppipe \
    --run-headless \
    --data-file ./test_image_set_list.csv \
    -o ./cp_out \
    -L 10

To run cellprofiler on slurm, first:

module load anaconda3
source activate cellprofiler-3.1.8

Merge single cell features calculated by cellprofiler and image metadata

To merge features, also need these files from repo:

1. fov metadata: data/input_segs_and_tiffs/labkey_fov_metadata.csv)
2. structure scores with fov id: data/structure_scores_fov.csv

merge_cellprofiler_output \
    --cp_csv_dir cp_out \
    --csv_prefix napari_ \
    --out_csv cp_out/cp_features.csv \
    --normalized_image_manifest quilt_data_contrasted_test/metadata.csv \
    --fov_metadata fish_morphology_code/data/input_segs_and_tiffs/labkey_fov_metadata.csv \
    --structure_scores fish_morphology_code/data/structure_scores_fov.csv \
    --prepend_local ./quilt_data_contrasted \
    --relative_columns "rescaled_2D_fov_tiff_path,rescaled_2D_single_cell_tiff_path"

Running the tests

To run the pytest tests defined in fish_morphology_code/tests via tox, use

make build

This will take a while the first time setting up the test environment.

Uploading data to quilt

Uploading to quilt should happen infrequently, so all the code has been put in the data directory outside of the main fish_morphology_code library. All uploading can be done in the data directory using python distribute_<your_dataset>.py.

Citation

@article {Gerbin2020.05.26.081083,
	author = {Gerbin, Kaytlyn A and Grancharova, Tanya and Donovan-Maiye, Rory and Hendershott, Melissa C and Brown, Jackson and Dinh, Stephanie Q and Gehring, Jamie L and Hirano, Matthew and Johnson, Gregory R and Nath, Aditya and Nelson, Angelique and Roco, Charles M and Rosenberg, Alex B and Sluzewski, M Filip and Viana, Matheus P and Yan, Calysta and Zaunbrecher, Rebecca J and Cordes Metzler, Kimberly R and Menon, Vilas and Palecek, Sean P and Seelig, Georg and Gaudreault, Nathalie and Knijnenburg, Theo and Rafelski, Susanne M and Theriot, Julie A and Gunawardane, Ruwanthi N},
	title = {Cell states beyond transcriptomics: integrating structural organization and gene expression in hiPSC-derived cardiomyocytes},
	elocation-id = {2020.05.26.081083},
	year = {2020},
	doi = {10.1101/2020.05.26.081083},
	publisher = {Cold Spring Harbor Laboratory},
	URL = {https://www.biorxiv.org/content/early/2020/05/27/2020.05.26.081083},
	eprint = {https://www.biorxiv.org/content/early/2020/05/27/2020.05.26.081083.full.pdf},
	journal = {bioRxiv}
}