diff --git a/MorphoMapping/__init__.py b/MorphoMapping/__init__.py
deleted file mode 100644
index 6dbfedd..0000000
--- a/MorphoMapping/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .morphomapping import MM
-
-__all__ = ["MM"]
diff --git a/MorphoMapping/morphomapping.py b/MorphoMapping/morphomapping.py
deleted file mode 100644
index 7399092..0000000
--- a/MorphoMapping/morphomapping.py
+++ /dev/null
@@ -1,1220 +0,0 @@
-# Author: Amelie Bauerdick
-# WabnitzLab
-# 2024
-
-
-"""
-Create interactive umap and densmap visualizations.
-Plot feature importance and apply several clustering algorithms after dimensionality reduction.
-Class MM is based on pandas DataFrames.
-
-Class:
-
- MM
-
-Functions:
-
- convert_to_CSV(fcs_path: str, csv_path: str)
- read_CSV(path: str, add_index: bool, index_name: str) -> df
- get_features -> list
- get_df -> df
- add_metadata(label: str, value) -> df
- rename_variables(label_mapping)
- select_condition(condition: str, value)
- select_events(event_size: int)
- drop_variables(*labels)
- drop_events(first_row: int, last_row: int)
- save_feature(*features) -> df
- concat_variables(*dataframes) -> df
- save_xlsx(path: str)
- save_csv(path: str)
- concat_df(*new_df, join='inner')
- update_column_values(column_name: str, rename_values)
- minmax_norm(first_column: str =None, last_column: str =None)
- quant_scaler(first_column: str =None, last_column: str =None)
- umap(nn: int, mdist: float, met: str)
- dmap(dlambda: float, nn: int, mdist: float, met: str)
- feature_importance(dep: str, indep: str) -> df
- plot_feature_importance(features, path: str, base_width: int, base_height: int)
- cluster_kmeans(n_cluster: int)
- cluster_gmm(number_component: int, random_s: int)
- cluster_hdbscan(cluster_size: int)
- check_dataframe()
- prepare_data_source()
- configure_hover_tooltips(feature: str, hover_tooltips: list[tuple[str, str]] = None)
- create_base_plot(fig_width: int, fig_height: int, fig_title: str, label_x: str, label_y: str, range_x: list[float], range_y: list[float], tools_emb, title_align: str)
- configure_axes_and_legend(plot, show_axes: bool, show_legend: bool)
- cat_plot(feature: str, subs: list[str], colors: list[str], outputf: str, fig_width: int, fig_height: int, fig_title: str, label_x: str,label_y: str, range_x: list[float], range_y: list[float],
- hover_tooltips: list[tuple[str, str]] = None, show_legend: bool = False, point_size: int, point_alpha: float, show_axes: bool = False, title_align: str = 'center') -> None:
- lin_plot(outputf: str, feature: str, colors: list[str], fig_width: int, fig_height: int, fig_title: str, label_x: str, label_y: str, range_x: list[float], range_y: list[float],
- hover_tooltips: list[tuple[str, str]] = None, show_legend: bool = False, point_size: int, point_alpha: float, show_axes: bool = True, title_align: str = 'center') -> None:
-
-variables:
-
- self.df
-
-"""
-
-
-
-# Import Packages
-
-import flowkit as fk
-import pandas as pd
-import numpy as np
-from matplotlib import pyplot as plt
-from sklearn.ensemble import RandomForestRegressor
-from sklearn.model_selection import train_test_split
-from sklearn.preprocessing import MinMaxScaler, QuantileTransformer
-from sklearn.metrics import r2_score
-import sklearn.cluster as cluster
-from sklearn.mixture import GaussianMixture
-import umap
-import hdbscan
-from bokeh.models import (HoverTool,
- ColumnDataSource,
- Range1d,
- LinearColorMapper,
- CategoricalColorMapper)
-from bokeh.plotting import figure, show, output_file
-import bokeh
-
-from typing import Optional
-
-bokeh_version = bokeh.__version__
-bokeh_above_three = int(bokeh_version.split(".")[0]) >= 3
-
-
-class MM:
-
- """\
- A class to create interactive dimensionality reduction plots.
- Based on pandas DataFrame.
-
- Parameters
- ----------
- df
- DataFrame containing all Imaging Flow Cytometry data
-
- Returns
- -------
- None
-
- """
-
-
-
- def __init__(self):
- self.df = pd.DataFrame()
-
- def convert_to_CSV(self, fcs_path: str, csv_path: str):
- """\
- Converts fcs-file to .csv file and saves it to csv_path.
-
- Parameters
- ----------
- fcs_path
- path to fcs file
- csv_path
- path to csv file
-
- Returns
- -------
- None
-
- """
- try:
- if not csv_path:
- raise ValueError("The CSV path is empty or invalid.")
-
- sample = fk.Sample(fcs_path)
- sample.export(filename=csv_path, source='raw')
-
- print(f"File successfully converted to {csv_path}.")
-
- except FileNotFoundError:
- print(f"FCS file not found: {fcs_path}")
- raise
- except Exception as e:
- print(f"Conversion failed: {e}")
- raise
-
- def read_CSV(self, path: str, add_index: bool =False, index_name: str ='Index'):
- """\
- Load csv-file and save it as self.df
-
- Parameters
- ----------
- path
- path to csv file
- add_index
- add index to df
- index_name
- set name of index
-
- Returns
- ----------
- DataFrame
-
- """
-
- self.df = pd.read_csv(path)
-
- #rename columns
- self.df.columns = (
- self.df.columns.str.strip()
- .str.replace(' ', '_')
- .str.replace('&', 'and')
- .str.replace('+', 'plus')
- .str.replace('-', 'minus')
- )
-
- if add_index:
- self.df[index_name] = range(len(self.df))
- self.df.set_index(index_name, inplace=True)
-
- return self.df
-
- def get_features(self):
- """\
- return list of self.df columns
- """
- return list(self.df.columns)
-
- def get_df(self):
- """\
- return self.df
- """
- return self.df
-
- def add_metadata(self, label: str, value):
- """\
- add column with specific value to self.df
-
- Parameters
- ----------
- label
- name of column
- value
- value of column
-
- Returns
- ----------
- DataFrame
-
- """
- empty = self.df.empty
- if empty:
- raise ValueError("Dataframe is empty.")
-
- self.df[label] = value
- return self.df
-
- def rename_variables(self, label_mapping):
- """\
- rename column(s) with new column labels
-
- Parameters
- ----------
- label_mapping
- old and new assigned column labels
-
- Returns
- ----------
- None
-
- """
-
- missing_columns = [col for col in label_mapping.keys() if col not in self.df.columns]
-
- if missing_columns:
- print(f"The following columns do not exist: {missing_columns}")
- else:
- self.df.rename(columns=label_mapping, inplace=True)
-
- def select_condition(self, condition: str, value):
- """\
- select specific rows by condition and save new df as self.df
-
- Parameters
- ----------
- condition
- name of column
- value
- value of column
-
- Returns
- ----------
- None
- """
-
- if condition not in self.df.columns:
- raise ValueError(f"Column '{condition}' does not exist.")
- if value not in self.df[condition].values:
- raise ValueError(f"Value '{value}' does not exist in column '{condition}'.")
- # select rows and save new df
- self.df = self.df.loc[self.df[condition] == value]
-
- def select_events(self, event_size: int):
- """\
- randomly select events and save as self.df
-
- Parameters
- ----------
- event_size
- number of events
-
- Returns
- ----------
- None
- """
-
- if event_size > self.df.shape[0]:
- raise ValueError(f"Number of events '{event_size}' is larger than the number of rows ({self.df.shape[0]}).")
- else:
- self.df = self.df.sample(event_size, random_state=1).copy()
- self.df.sort_index(inplace=True)
-
- def drop_variables(self, *labels):
- """\
- drop certain columns from self.df
-
- Parameters
- ----------
- labels
- name of column(s)
-
- Returns
- ----------
- None
- """
- missing_labels = [label for label in labels if label not in self.df.columns]
- if missing_labels:
- raise ValueError(f"Column(s) {missing_labels} do not exist.")
- else:
- self.df = self.df.drop(columns=list(labels))
-
- def drop_events(self, first_row: int, last_row: int):
- """\
- drop specific rows from self.df
-
- Parameters
- ----------
- first_row
- number of first row that should be dropped
- last_row
- number of last row that should be dropped
-
- Returns
- ----------
- None
- """
- if first_row not in self.df.index:
- raise ValueError(f"First row '{first_row}' does not exist.")
- if last_row not in self.df.index:
- raise ValueError(f"Last row '{last_row}' does not exist.")
- if first_row > last_row:
- raise ValueError(f"First row number '{first_row}' is greater than last row number '{last_row}'.")
- self.df = self.df.drop(index=self.df.index[first_row:last_row + 1])
-
- def save_feature(self, *features):
- """\
- save specific columns of self.df in new DataFrame and return new DataFrame
-
- Parameters
- ----------
- features
- name of features which should be returned as df
-
- Returns
- ----------
- DataFrame
- """
- labels = [feature for feature in features if feature not in self.df.columns]
- if labels:
- raise ValueError(f"Column(s) {labels} do not exist.")
-
- df = self.df[list(features)].copy()
-
- return df
-
- def concat_variables(self, *dataframes):
- """\
- attach new columns as df to self.df and return resulting df
-
- Parameters
- ----------
- dataframes
- new dataframes which should be attached to self.df
-
- Returns
- ----------
- DataFrame
- """
- result_df = pd.concat([self.df] + list(dataframes), axis=1)
- self.df = result_df
- return result_df
-
- def save_xlsx(self, path: str):
- """\
- save self.df as xlsx file to chosen path
-
- Parameters
- ----------
- path
- chosen path
-
- Returns
- ----------
- None
- """
- self.df.to_excel(path, index=False)
- print(f"DataFrame successfully saved to {path}")
-
- def save_csv(self, path: str):
- """\
- save self.df as csv to chosen path
-
- Parameters
- ----------
- path
- chosen path
-
- Returns
- ----------
- None
- """
- self.df.to_csv(path, index=False)
- print(f"DataFrame successfully saved to {path}")
-
- def concat_df(self, *new_df, join='inner'):
- """\
- concatenate self.df and new DataFrame(s) (joining inner by default)
-
- Parameters
- ----------
- new_df
- DataFrames that should be concatenated
- join
- type of joining DataFrames
-
- Returns
- ----------
- None
- """
- self.df = pd.concat([self.df, *new_df], join=join)
-
- def update_column_values(self, column_name: str, rename_values):
- """\
- replace values in a specific column with a specific new value
-
- Parameters
- ----------
- column_name
- name of column
- rename_values
- new value of column
-
- Returns
- ----------
- None
- """
-
- if column_name not in self.df.columns:
- raise ValueError(f"Column '{column_name}' does not exist.")
-
- self.df[column_name] = self.df[column_name].astype(str)
-
- for original_value, new_value in rename_values.items():
- self.df[column_name] = self.df[column_name].str.replace(str(original_value), new_value)
-
- def minmax_norm(self, first_column: Optional[str] = None, last_column: str =None):
- """\
- Apply MinMax normalization to self.df.
- Specify whether all columns should be normalized by setting parameters.
-
- Parameters
- ----------
- first_column
- name of first column. Start of normalization.
- last_column:
- name of last column. End of normalization.
-
- Returns
- ----------
- None
-
- """
-
- #check df
- if first_column is None or last_column is None:
- df1 = self.df
- else:
- if first_column not in self.df.columns:
- raise ValueError(f"First column '{first_column}' does not exist.")
- if last_column not in self.df.columns:
- raise ValueError(f"Last column '{last_column}' does not exist.")
-
- first_id = self.df.columns.get_loc(first_column)
- last_id = self.df.columns.get_loc(last_column)
- if first_id > last_id:
- raise ValueError(f"First column '{first_column}' is after last column '{last_column}'.")
-
- df1 = self.df.iloc[:, first_id:last_id + 1]
-
- # Apply min-max normalization
- df = (df1 - df1.min()) / (df1.max() - df1.min())
-
- # Replace the subset
- if first_column is not None and last_column is not None:
- self.df.iloc[:, first_id:last_id + 1] = df
- else:
- self.df = df
-
-
- def quant_scaler(self, first_column: Optional[str] = None, last_column: str =None):
- """\
- Apply QuantileTransformer to self.df.
- Specify whether all columns should be normalized by setting parameters.
-
- Parameters
- ----------
- first_column
- name of first column. Start of normalization.
- last_column
- name of last column. End of normalization.
-
- Returns
- ----------
- None
- """
-
- if first_column is None or last_column is None:
- df1 = self.df
- else:
- if first_column not in self.df.columns:
- raise ValueError(f"First column '{first_column}' does not exist.")
- if last_column not in self.df.columns:
- raise ValueError(f"Last column '{last_column}' does not exist.")
-
- first_id = self.df.columns.get_loc(first_column)
- last_id = self.df.columns.get_loc(last_column)
- if first_id > last_id:
- raise ValueError(f"First column '{first_column}' is after last column '{last_column}'.")
-
- df1 = self.df.iloc[:, first_id:last_id + 1]
-
- # Apply Quantile Transformation
- scaler = QuantileTransformer(output_distribution='uniform')
- df2 = pd.DataFrame(scaler.fit_transform(df1),
- columns=df1.columns,
- index=df1.index)
-
- if first_column is not None and last_column is not None:
- self.df.iloc[:, first_id:last_id + 1] = df2
- else:
- self.df = df2
-
-
- def umap(self, nn: int, mdist: float, met: str):
- """\
- Run umap with self.df. Adds x and y values to self.df as extra columns.
-
- Parameters
- ----------
- nn
- nearest neighbours for umap settings
- mdist
- minimum distance for umap settings
- met
- metric for umap settings
-
- Returns
- ----------
- None
-
- """
- reducer = umap.UMAP(
- n_neighbors=nn,
- min_dist=mdist,
- metric=met
- )
-
- embedding = reducer.fit_transform(self.df)
-
- x = embedding[:, 0]
- y = embedding[:, 1]
-
- self.df['x'] = x
- self.df['y'] = y
-
- def dmap(self, dlambda: float, nn: int, mdist: float, met: str):
- """\
- Run dmap with self.df. Adds x and y values to self.df as extra columns.
-
- Parameters
- ----------
- dlambda
- denslambda for densmap settings
- nn
- nearest neighbours for densmap settings
- mdist
- minimum distance for densmap settings
- met
- metric for densmap settings
-
- Returns
- ----------
- None
- """
-
- reducer = umap.UMAP(
- densmap=True,
- dens_lambda=dlambda,
- n_neighbors=nn,
- min_dist=mdist,
- metric=met
- )
-
- embedding = reducer.fit_transform(self.df)
- x = embedding[:, 0]
- y = embedding[:, 1]
-
- self.df['x'] = x
- self.df['y'] = y
-
- def feature_importance(self, dep: str, indep: str):
- """\
- Calculates feature importance of columns in self.df (especially for x and y after dmap/umap were run).
- Returns DataFrame with the 10 most important features and their according importance values.
-
- Parameters
- ----------
- dep
- name of column which should represent the dependent variable
- indep
- name of column which should represent the independent variable
-
- Returns
- ----------
- DataFrame
-
- """
-
- data = self.df.copy()
- data = data.drop(indep, axis=1)
-
- #split data
- train_df, test_df = train_test_split(data, test_size=0.2, random_state=42)
-
- X_train = train_df.drop(dep, axis=1)
- y_train = train_df[dep]
- X_test = test_df.drop(dep, axis=1)
- y_test = test_df[dep]
-
- print("length of data for training:", len(X_train))
- print("length of data for testing:", len(X_test))
-
- # run RandomForestRegressor
- model = RandomForestRegressor(n_estimators=100, random_state=42)
- model.fit(X_train, y_train)
-
- # predict dependent variable
- y_pred = model.predict(X_test)
-
- # r²-value calculation
- r2 = r2_score(y_test, y_pred)
- print("r² Score:", r2)
-
- # Feature Importance
- importance = model.feature_importances_
-
- # sort features according to importance
- s_id = np.argsort(importance)
- pos = np.arange(s_id.shape[0])
-
- # MinMax scaling
- scaler = MinMaxScaler()
- importance_scaled = scaler.fit_transform(importance.reshape(-1, 1)).flatten()
-
- # importance
- total_importance = np.sum(importance_scaled)
- percentage_importance = (importance_scaled / total_importance) * 100
-
- # show top ten
- top_n = 10
- s_id = s_id[-top_n:]
- features = pd.DataFrame(
- {'index1': np.array(X_train.columns)[s_id], 'importance_normalized': importance_scaled[s_id],
- 'percentage_importance': percentage_importance[s_id]})
- return features
-
-
- def plot_feature_importance(self, features, path: str, base_width: int =10, base_height: int =6):
- """\
- Plots the ten most important features and returns a pyplot.
- Needs the ten top features and their importances as parameters.
-
- Parameters
- ----------
- features
- DataFrame returned by function feature_importance
- path
- path to dict where plot should be saved
- base_width
- width of plot
- base_height
- height of plot
-
- Returns
- ----------
- Show plot
-
- """
-
- num_features = len(features)
-
- plot_width = base_width
- plot_height = base_height + 0.2 * num_features
-
- ax = features.plot.bar(x='index1',
- y='importance_normalized',
- color='darkgray',
- legend=False,
- figsize=(plot_width, plot_height),
- width=0.8, fontsize=20)
-
- plt.xlabel('')
- plt.ylabel('Importance', fontsize=20)
-
- #adjust text height
- for i, v in enumerate(features['percentage_importance']):
- if features['importance_normalized'][i] + 0.01 > 1.1:
- text_height = 1
- else:
- text_height = features['importance_normalized'][i] + 0.01
- ax.text(i, text_height, f'{v:.1f}%', ha='center', va='bottom', fontsize=16, color='black')
-
- plt.title('Top 10 Features', fontsize=30, loc='left')
- plt.ylim(0, 1.1)
- plt.xticks(rotation=45, ha='right')
-
- if path is not None:
- try:
- plt.savefig(path, dpi=300, bbox_inches='tight')
- print(f"Plot successfully saved to {path}")
- except Exception as e:
- print(f"An error occurred while saving to png: {e}")
-
- return plt.show()
-
- def cluster_kmeans(self, n_cluster: int, label_x: str, label_y: str):
- """\
- Cluster self.df by kmeans clustering and show result as plt.show().
-
- Parameters
- ----------
- number_cluster
- number of clusters
- label_x
- x-axis label
- label_y
- y-axis label
-
- Returns
- ----------
- Show plot
- """
-
- kmeans_labels = cluster.KMeans(n_clusters=n_cluster).fit_predict(self.df)
-
- #plot
- plt.style.use('seaborn-v0_8-poster')
- plt.figure(figsize=(6, 6))
-
- plt.scatter(self.df[['x']],
- self.df[['y']],
- c=kmeans_labels,
- s=1,
- cmap='Set1');
-
- plt.title('K-Means Clustering')
- plt.xlabel(label_x)
- plt.ylabel(label_y)
- plt.xticks([])
- plt.yticks([])
-
- self.df['kmeans_cluster'] = kmeans_labels.tolist()
- return plt.show()
-
- def cluster_gmm(self, number_component: int, random_s: int, label_x: str, label_y: str):
- """\
- Cluster self.df by Gaussian Mixture Modeles and plot result.
-
- Parameters
- ----------
- number_component
- number of components of gmm clustering
- random_s
- random state of gmm clustering
- label_x
- x-axis label
- label_y
- y-axis label
-
- Returns
- ----------
- Show plot
-
- """
-
- gmm = GaussianMixture(n_components=number_component, random_state=random_s)
-
- gmm.fit(self.df)
- gaussian_labels = gmm.predict(self.df)
-
- #plot
- plt.style.use('seaborn-v0_8-poster')
- plt.figure(figsize=(6, 6))
-
- plt.scatter(self.df[['x']],
- self.df[['y']],
- c=gaussian_labels,
- s=1,
- cmap='Set1');
-
- plt.title('Gaussian Mixture Clustering')
- plt.xlabel(label_x)
- plt.ylabel(label_y)
- plt.xticks([])
- plt.yticks([])
-
- self.df['GMM_cluster'] = gaussian_labels.tolist()
- return plt.show()
-
- def cluster_hdbscan(self, cluster_size: int, label_x: str, label_y: str):
- """\
- Cluster self.df by hdbscan and plot result.
-
- Parameters
- ----------
- cluster_size
- size of clusters
- label_x
- x-axis label
- label_y
- y-axis label
-
- Returns
- ----------
- Show plot
-
- """
-
- clusterer = hdbscan.HDBSCAN(min_cluster_size=cluster_size, gen_min_span_tree=True)
-
- clusterer.fit(self.df)
- hdbscan_labels = clusterer.labels_
-
- # Plot
- outliers_mask = hdbscan_labels == -1
- plt.style.use('seaborn-v0_8-poster')
- plt.figure(figsize=(6, 6))
-
- plt.scatter(self.df[['x']],
- self.df[['y']],
- c=hdbscan_labels,
- cmap='Spectral',
- s=5)
-
- plt.scatter(self.df.loc[outliers_mask, 'x'],
- self.df.loc[outliers_mask, 'y'],
- s=4,
- c='gray',
- marker='v',
- label='Outliers',
- alpha=0.5)
-
- plt.title('HDBSCAN Clustering')
- plt.xlabel(label_x)
- plt.ylabel(label_y)
- plt.legend(markerscale=6)
-
- plt.xticks([])
- plt.yticks([])
-
- self.df['hdbscan_cluster'] = hdbscan_labels.tolist()
- return plt.show()
-
-
-
- #def for dmap/umap plots
-
- def check_dataframe(self):
- """\
- check if df is empty
- """
- if self.df.empty:
- raise ValueError("Dataframe is empty.")
-
- def prepare_data_source(self):
- """\
- set ColumnDataSource as self.df and return ColumnDataSource
- """
- return ColumnDataSource(self.df)
-
- def configure_hover_tooltips(self, feature: str, hover_tooltips: list[tuple[str, str]] = None):
- """\
- Set hover tooltips. Either standard or personalized.
-
- Parameters
- ----------
- feature
- column of dataset
- hover_tooltips
-
- Returns
- ----------
- HoverTool
-
- """
- if hover_tooltips is None:
- hover_tooltips = [
- ("Feature", f"{feature}"),
- ("Index", "ID"),
- ("X-Value", "x"),
- ("Y-Value", "y")
- ]
-
- hover_tooltips_formatted = "".join([
- f"{label}: @{field}
"
- for label, field in hover_tooltips
- ])
-
- hovertool_kwargs = {
- "name": "data",
- "tooltips": hover_tooltips_formatted
- }
- return HoverTool(**hovertool_kwargs)
-
- def create_base_plot(self, fig_width: int,
- fig_height: int,
- fig_title: str,
- label_x: str,
- label_y: str,
- range_x: list[float],
- range_y: list[float],
- tools_emb,
- title_align: str):
- """\
- Settings for plotting umap/densmap.
-
- Parameters
- ----------
- fig_width
- width of figure
- fig_height
- height of figure
- fig_title
- title of figure
- label_x
- x-axis label
- label_y
- y-axis label
- range_x
- range of x-axis
- range_y
- range of y-axis
- tools_emb
- tools that should be embedded
- title_align
- position of title
-
- Returns
- ----------
- plot
-
- """
- if not bokeh_above_three:
- plot_kwargs = dict(
- plot_width=fig_width,
- plot_height=fig_height,
- title=fig_title,
- tools=tools_emb,
- x_axis_label=label_x,
- y_axis_label=label_y,
- x_range=Range1d(start=range_x[0], end=range_x[1]),
- y_range=Range1d(start=range_y[0], end=range_y[1])
- )
- else:
- plot_kwargs = dict(
- width=fig_width,
- height=fig_height,
- title=fig_title,
- tools=tools_emb,
- x_axis_label=label_x,
- y_axis_label=label_y,
- x_range=Range1d(start=range_x[0], end=range_x[1]),
- y_range=Range1d(start=range_y[0], end=range_y[1])
- )
-
- plot = figure(**plot_kwargs)
-
- # configure title
- plot.title.align = title_align
- plot.title.text_font_size = '30pt'
-
- return plot
-
- def configure_axes_and_legend(self, plot, show_axes: bool, show_legend: bool):
- """\
- Settings for axes and legend.
-
- Parameters
- ----------
- plot
- show_axes
- add axes
- show_legend
- add legend
-
- Returns
- ----------
- None
-
- """
-
- #axes
- if not show_axes:
- plot.xaxis.visible = False
- plot.yaxis.visible = False
-
- plot.grid.visible = False
- plot.outline_line_color = None
-
- plot.xaxis.axis_label_text_font_size = "20pt"
- plot.yaxis.axis_label_text_font_size = "20pt"
-
- plot.xaxis.ticker = []
- plot.yaxis.ticker = []
-
- # legend
- if show_legend:
- plot.legend.title_text_font_style = "bold"
- plot.legend.background_fill_alpha = 0.0
- plot.legend.border_line_alpha = 0
- plot.legend.label_text_font_size = '20pt'
- plot.legend.title_text_font_size = '20pt'
- plot.legend.glyph_height = 30
- plot.legend.glyph_width = 30
- plot.add_layout(plot.legend[0], 'center')
- else:
- plot.legend.visible = False
-
-
- def cat_plot(self, feature: str,
- subs: list[str],
- colors: list[str],
- outputf: str,
- fig_width: int,
- fig_height: int,
- fig_title: str,
- label_x: str,
- label_y: str,
- range_x: list[float],
- range_y: list[float],
- hover_tooltips: list[tuple[str, str]] = None,
- show_legend: bool = False,
- point_size: int = 10,
- point_alpha: float = 0.6,
- show_axes: bool = False,
- title_align: str = 'center') -> None:
-
- """\
- Create plot with categorical color mapper.Choose feature, colors and more.
- Loads html file.
-
- Parameters
- ----------
- feature
- feature for categorical mapper
- subs
- values of feature
- colors
- list of colors
- outputf
- path to outputfile
- fig_width
- width of figure
- fig_height
- height of figure
- fig_title
- title of figure
- label_x
- x-axis label
- label_y
- y-axis label
- range_x
- x-axis range
- range_y
- y-axis range
- hover_tooltips
- hover tooltips
- show_legend
- possibility to add legend
- point_size
- size of points
- point_alpha
- alpha of points
- show_axes
- add axes
- title_align
- position of title
-
- Returns
- ----------
- show plot
-
- """
-
- self.check_dataframe()
- output_file(outputf)
-
- source = self.prepare_data_source()
- hover_emb = self.configure_hover_tooltips(feature, hover_tooltips)
- cm = CategoricalColorMapper(palette=colors, factors=subs)
-
- tools_emb = ['save', 'lasso_select', 'pan', 'wheel_zoom', 'reset', hover_emb]
- plot = self.create_base_plot(fig_width,
- fig_height,
- fig_title,
- label_x,
- label_y,
- range_x,
- range_y,
- tools_emb,
- title_align)
-
- plot.circle('x', 'y',
- size=point_size,
- color={'field': feature, 'transform': cm},
- alpha=point_alpha,
- source=source,
- name="data",
- legend_group=feature)
-
- self.configure_axes_and_legend(plot, show_axes, show_legend)
-
- show(plot)
-
- def lin_plot(self,
- outputf: str,
- feature: str,
- colors: list[str],
- fig_width: int,
- fig_height: int,
- fig_title: str,
- label_x: str,
- label_y: str,
- range_x: list[float],
- range_y: list[float],
- hover_tooltips: list[tuple[str, str]] = None,
- show_legend: bool = False,
- point_size: int = 3,
- point_alpha: float = 0.7,
- show_axes: bool = True,
- title_align: str = 'center') -> None:
-
-
- """\
- Create plot with Linear color mapper.Choose feature, colors and more.
- Loads html file.
-
- Parameters
- ----------
- outputf
- path to output file
- feature
- feature for linear mapper
- colors
- list of colors
- fig_width
- width of figure
- fig_height
- height of figure
- fig_title
- title of figure
- label_x
- x-axis label
- label_y
- y-axis label
- range_x
- range of x-axis
- range_y
- range of y-axis
- hover_tooltips
- hover tooltips
- show_legend
- add legend
- point_size
- size of points
- point_alpha
- alpha of points
- show_axes
- add axes
- title_align
- title position
-
- Returns
- ----------
- show plot
-
- """
-
-
- self.check_dataframe()
- output_file(outputf)
-
- source = self.prepare_data_source()
- hover_emb = self.configure_hover_tooltips(feature, hover_tooltips)
- lm = LinearColorMapper(palette=colors, low=min(self.df[feature]), high=max(self.df[feature]))
-
- tools_emb = ['save', 'lasso_select', 'pan', 'wheel_zoom', 'reset', hover_emb]
- plot = self.create_base_plot(fig_width,
- fig_height,
- fig_title,
- label_x,
- label_y,
- range_x,
- range_y,
- tools_emb,
- title_align)
-
- plot.circle('x', 'y',
- size=point_size,
- fill_color={'field': feature, 'transform': lm},
- alpha=point_alpha,
- line_alpha=0,
- line_width=0.03,
- source=source,
- name="data",
- legend_group=feature)
-
- self.configure_axes_and_legend(plot, show_axes, show_legend)
-
- show(plot)
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
diff --git a/morphomapping/__init__.py b/morphomapping/__init__.py
deleted file mode 100644
index 6dbfedd..0000000
--- a/morphomapping/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .morphomapping import MM
-
-__all__ = ["MM"]
diff --git a/morphomapping/morphomapping.py b/morphomapping/morphomapping.py
deleted file mode 100644
index 493d7af..0000000
--- a/morphomapping/morphomapping.py
+++ /dev/null
@@ -1,1220 +0,0 @@
-# Author: Amelie Bauerdick
-# WabnitzLab
-# 2024
-
-
-"""
-Create interactive umap and densmap visualizations.
-Plot feature importance and apply several clustering algorithms after dimensionality reduction.
-Class MM is based on pandas DataFrames.
-
-Class:
-
- MM
-
-Functions:
-
- convert_to_CSV(fcs_path: str, csv_path: str)
- read_CSV(path: str, add_index: bool, index_name: str) -> df
- get_features -> list
- get_df -> df
- add_metadata(label: str, value) -> df
- rename_variables(label_mapping)
- select_condition(condition: str, value)
- select_events(event_size: int)
- drop_variables(*labels)
- drop_events(first_row: int, last_row: int)
- save_feature(*features) -> df
- concat_variables(*dataframes) -> df
- save_xlsx(path: str)
- save_csv(path: str)
- concat_df(*new_df, join='inner')
- update_column_values(column_name: str, rename_values)
- minmax_norm(first_column: str =None, last_column: str =None)
- quant_scaler(first_column: str =None, last_column: str =None)
- umap(nn: int, mdist: float, met: str)
- dmap(dlambda: float, nn: int, mdist: float, met: str)
- feature_importance(dep: str, indep: str) -> df
- plot_feature_importance(features, path: str, base_width: int, base_height: int)
- cluster_kmeans(n_cluster: int)
- cluster_gmm(number_component: int, random_s: int)
- cluster_hdbscan(cluster_size: int)
- check_dataframe()
- prepare_data_source()
- configure_hover_tooltips(feature: str, hover_tooltips: list[tuple[str, str]] = None)
- create_base_plot(fig_width: int, fig_height: int, fig_title: str, label_x: str, label_y: str, range_x: list[float], range_y: list[float], tools_emb, title_align: str)
- configure_axes_and_legend(plot, show_axes: bool, show_legend: bool)
- cat_plot(feature: str, subs: list[str], colors: list[str], outputf: str, fig_width: int, fig_height: int, fig_title: str, label_x: str,label_y: str, range_x: list[float], range_y: list[float],
- hover_tooltips: list[tuple[str, str]] = None, show_legend: bool = False, point_size: int, point_alpha: float, show_axes: bool = False, title_align: str = 'center') -> None:
- lin_plot(outputf: str, feature: str, colors: list[str], fig_width: int, fig_height: int, fig_title: str, label_x: str, label_y: str, range_x: list[float], range_y: list[float],
- hover_tooltips: list[tuple[str, str]] = None, show_legend: bool = False, point_size: int, point_alpha: float, show_axes: bool = True, title_align: str = 'center') -> None:
-
-variables:
-
- self.df
-
-"""
-
-
-
-# Import Packages
-
-import flowkit as fk
-import pandas as pd
-import numpy as np
-from matplotlib import pyplot as plt
-from sklearn.ensemble import RandomForestRegressor
-from sklearn.model_selection import train_test_split
-from sklearn.preprocessing import MinMaxScaler, QuantileTransformer
-from sklearn.metrics import r2_score
-import sklearn.cluster as cluster
-from sklearn.mixture import GaussianMixture
-import umap
-import hdbscan
-from bokeh.models import (HoverTool,
- ColumnDataSource,
- Range1d,
- LinearColorMapper,
- CategoricalColorMapper)
-from bokeh.plotting import figure, show, output_file
-import bokeh
-
-from typing import Optional
-
-bokeh_version = bokeh.__version__
-bokeh_above_three = int(bokeh_version.split(".")[0]) >= 3
-
-
-class MM:
-
- """\
- A class to create interactive dimensionality reduction plots.
- Based on pandas DataFrame.
-
- Parameters
- ----------
- df
- DataFrame containing all Imaging Flow Cytometry data
-
- Returns
- -------
- None
-
- """
-
-
-
- def __init__(self):
- self.df = pd.DataFrame()
-
- def convert_to_CSV(self, fcs_path: str, csv_path: str):
- """\
- Converts fcs-file to .csv file and saves it to csv_path.
-
- Parameters
- ----------
- fcs_path
- path to fcs file
- csv_path
- path to csv file
-
- Returns
- -------
- None
-
- """
- try:
- if not csv_path:
- raise ValueError("The CSV path is empty or invalid.")
-
- sample = fk.Sample(fcs_path)
- sample.export(filename=csv_path, source='raw')
-
- print(f"File successfully converted to {csv_path}.")
-
- except FileNotFoundError:
- print(f"FCS file not found: {fcs_path}")
- raise
- except Exception as e:
- print(f"Conversion failed: {e}")
- raise
-
- def read_CSV(self, path: str, add_index: bool =False, index_name: str ='Index'):
- """\
- Load csv-file and save it as self.df
-
- Parameters
- ----------
- path
- path to csv file
- add_index
- add index to df
- index_name
- set name of index
-
- Returns
- ----------
- DataFrame
-
- """
-
- self.df = pd.read_csv(path)
-
- #rename columns
- self.df.columns = (
- self.df.columns.str.strip()
- .str.replace(' ', '_')
- .str.replace('&', 'and')
- .str.replace('+', 'plus')
- .str.replace('-', 'minus')
- )
-
- if add_index:
- self.df[index_name] = range(len(self.df))
- self.df.set_index(index_name, inplace=True)
-
- return self.df
-
- def get_features(self):
- """\
- return list of self.df columns
- """
- return list(self.df.columns)
-
- def get_df(self):
- """\
- return self.df
- """
- return self.df
-
- def add_metadata(self, label: str, value):
- """\
- add column with specific value to self.df
-
- Parameters
- ----------
- label
- name of column
- value
- value of column
-
- Returns
- ----------
- DataFrame
-
- """
- empty = self.df.empty
- if empty:
- raise ValueError("Dataframe is empty.")
-
- self.df[label] = value
- return self.df
-
- def rename_variables(self, label_mapping):
- """\
- rename column(s) with new column labels
-
- Parameters
- ----------
- label_mapping
- old and new assigned column labels
-
- Returns
- ----------
- None
-
- """
-
- missing_columns = [col for col in label_mapping.keys() if col not in self.df.columns]
-
- if missing_columns:
- print(f"The following columns do not exist: {missing_columns}")
- else:
- self.df.rename(columns=label_mapping, inplace=True)
-
- def select_condition(self, condition: str, value):
- """\
- select specific rows by condition and save new df as self.df
-
- Parameters
- ----------
- condition
- name of column
- value
- value of column
-
- Returns
- ----------
- None
- """
-
- if condition not in self.df.columns:
- raise ValueError(f"Column '{condition}' does not exist.")
- if value not in self.df[condition].values:
- raise ValueError(f"Value '{value}' does not exist in column '{condition}'.")
- # select rows and save new df
- self.df = self.df.loc[self.df[condition] == value]
-
- def select_events(self, event_size: int):
- """\
- randomly select events and save as self.df
-
- Parameters
- ----------
- event_size
- number of events
-
- Returns
- ----------
- None
- """
-
- if event_size > self.df.shape[0]:
- raise ValueError(f"Number of events '{event_size}' is larger than the number of rows ({self.df.shape[0]}).")
- else:
- self.df = self.df.sample(event_size, random_state=1).copy()
- self.df.sort_index(inplace=True)
-
- def drop_variables(self, *labels):
- """\
- drop certain columns from self.df
-
- Parameters
- ----------
- labels
- name of column(s)
-
- Returns
- ----------
- None
- """
- missing_labels = [label for label in labels if label not in self.df.columns]
- if missing_labels:
- raise ValueError(f"Column(s) {missing_labels} do not exist.")
- else:
- self.df = self.df.drop(columns=list(labels))
-
- def drop_events(self, first_row: int, last_row: int):
- """\
- drop specific rows from self.df
-
- Parameters
- ----------
- first_row
- number of first row that should be dropped
- last_row
- number of last row that should be dropped
-
- Returns
- ----------
- None
- """
- if first_row not in self.df.index:
- raise ValueError(f"First row '{first_row}' does not exist.")
- if last_row not in self.df.index:
- raise ValueError(f"Last row '{last_row}' does not exist.")
- if first_row > last_row:
- raise ValueError(f"First row number '{first_row}' is greater than last row number '{last_row}'.")
- self.df = self.df.drop(index=self.df.index[first_row:last_row + 1])
-
- def save_feature(self, *features):
- """\
- save specific columns of self.df in new DataFrame and return new DataFrame
-
- Parameters
- ----------
- features
- name of features which should be returned as df
-
- Returns
- ----------
- DataFrame
- """
- labels = [feature for feature in features if feature not in self.df.columns]
- if labels:
- raise ValueError(f"Column(s) {labels} do not exist.")
-
- df = self.df[list(features)].copy()
-
- return df
-
- def concat_variables(self, *dataframes):
- """\
- attach new columns as df to self.df and return resulting df
-
- Parameters
- ----------
- dataframes
- new dataframes which should be attached to self.df
-
- Returns
- ----------
- DataFrame
- """
- result_df = pd.concat([self.df] + list(dataframes), axis=1)
- self.df = result_df
- return result_df
-
- def save_xlsx(self, path: str):
- """\
- save self.df as xlsx file to chosen path
-
- Parameters
- ----------
- path
- chosen path
-
- Returns
- ----------
- None
- """
- self.df.to_excel(path, index=False)
- print(f"DataFrame successfully saved to {path}")
-
- def save_csv(self, path: str):
- """\
- save self.df as csv to chosen path
-
- Parameters
- ----------
- path
- chosen path
-
- Returns
- ----------
- None
- """
- self.df.to_csv(path, index=False)
- print(f"DataFrame successfully saved to {path}")
-
- def concat_df(self, *new_df, join='inner'):
- """\
- concatenate self.df and new DataFrame(s) (joining inner by default)
-
- Parameters
- ----------
- new_df
- DataFrames that should be concatenated
- join
- type of joining DataFrames
-
- Returns
- ----------
- None
- """
- self.df = pd.concat([self.df, *new_df], join=join)
-
- def update_column_values(self, column_name: str, rename_values):
- """\
- replace values in a specific column with a specific new value
-
- Parameters
- ----------
- column_name
- name of column
- rename_values
- new value of column
-
- Returns
- ----------
- None
- """
-
- if column_name not in self.df.columns:
- raise ValueError(f"Column '{column_name}' does not exist.")
-
- self.df[column_name] = self.df[column_name].astype(str)
-
- for original_value, new_value in rename_values.items():
- self.df[column_name] = self.df[column_name].str.replace(str(original_value), new_value)
-
- def minmax_norm(self, first_column: Optional[str] = None, last_column: str =None):
- """\
- Apply MinMax normalization to self.df.
- Specify whether all columns should be normalized by setting parameters.
-
- Parameters
- ----------
- first_column
- name of first column. Start of normalization.
- last_column:
- name of last column. End of normalization.
-
- Returns
- ----------
- None
-
- """
-
- #check df
- if first_column is None or last_column is None:
- df1 = self.df
- else:
- if first_column not in self.df.columns:
- raise ValueError(f"First column '{first_column}' does not exist.")
- if last_column not in self.df.columns:
- raise ValueError(f"Last column '{last_column}' does not exist.")
-
- first_id = self.df.columns.get_loc(first_column)
- last_id = self.df.columns.get_loc(last_column)
- if first_id > last_id:
- raise ValueError(f"First column '{first_column}' is after last column '{last_column}'.")
-
- df1 = self.df.iloc[:, first_id:last_id + 1]
-
- # Apply min-max normalization
- df = (df1 - df1.min()) / (df1.max() - df1.min())
-
- # Replace the subset
- if first_column is not None and last_column is not None:
- self.df.iloc[:, first_id:last_id + 1] = df
- else:
- self.df = df
-
-
- def quant_scaler(self, first_column: Optional[str] = None, last_column: str =None):
- """\
- Apply QuantileTransformer to self.df.
- Specify whether all columns should be normalized by setting parameters.
-
- Parameters
- ----------
- first_column
- name of first column. Start of normalization.
- last_column
- name of last column. End of normalization.
-
- Returns
- ----------
- None
- """
-
- if first_column is None or last_column is None:
- df1 = self.df
- else:
- if first_column not in self.df.columns:
- raise ValueError(f"First column '{first_column}' does not exist.")
- if last_column not in self.df.columns:
- raise ValueError(f"Last column '{last_column}' does not exist.")
-
- first_id = self.df.columns.get_loc(first_column)
- last_id = self.df.columns.get_loc(last_column)
- if first_id > last_id:
- raise ValueError(f"First column '{first_column}' is after last column '{last_column}'.")
-
- df1 = self.df.iloc[:, first_id:last_id + 1]
-
- # Apply Quantile Transformation
- scaler = QuantileTransformer(output_distribution='uniform')
- df2 = pd.DataFrame(scaler.fit_transform(df1),
- columns=df1.columns,
- index=df1.index)
-
- if first_column is not None and last_column is not None:
- self.df.iloc[:, first_id:last_id + 1] = df2
- else:
- self.df = df2
-
-
- def umap(self, nn: int, mdist: float, met: str):
- """\
- Run umap with self.df. Adds x and y values to self.df as extra columns.
-
- Parameters
- ----------
- nn
- nearest neighbours for umap settings
- mdist
- minimum distance for umap settings
- met
- metric for umap settings
-
- Returns
- ----------
- None
-
- """
- reducer = umap.UMAP(
- n_neighbors=nn,
- min_dist=mdist,
- metric=met
- )
-
- embedding = reducer.fit_transform(self.df)
-
- x = embedding[:, 0]
- y = embedding[:, 1]
-
- self.df['x'] = x
- self.df['y'] = y
-
- def dmap(self, dlambda: float, nn: int, mdist: float, met: str):
- """\
- Run dmap with self.df. Adds x and y values to self.df as extra columns.
-
- Parameters
- ----------
- dlambda
- denslambda for densmap settings
- nn
- nearest neighbours for densmap settings
- mdist
- minimum distance for densmap settings
- met
- metric for densmap settings
-
- Returns
- ----------
- None
- """
-
- reducer = umap.UMAP(
- densmap=True,
- dens_lambda=dlambda,
- n_neighbors=nn,
- min_dist=mdist,
- metric=met
- )
-
- embedding = reducer.fit_transform(self.df)
- x = embedding[:, 0]
- y = embedding[:, 1]
-
- self.df['x'] = x
- self.df['y'] = y
-
- def feature_importance(self, dep: str, indep: str):
- """\
- Calculates feature importance of columns in self.df (especially for x and y after dmap/umap were run).
- Returns DataFrame with the 10 most important features and their according importance values.
-
- Parameters
- ----------
- dep
- name of column which should represent the dependent variable
- indep
- name of column which should represent the independent variable
-
- Returns
- ----------
- DataFrame
-
- """
-
- data = self.df.copy()
- data = data.drop(indep, axis=1)
-
- #split data
- train_df, test_df = train_test_split(data, test_size=0.2, random_state=42)
-
- X_train = train_df.drop(dep, axis=1)
- y_train = train_df[dep]
- X_test = test_df.drop(dep, axis=1)
- y_test = test_df[dep]
-
- print("length of data for training:", len(X_train))
- print("length of data for testing:", len(X_test))
-
- # run RandomForestRegressor
- model = RandomForestRegressor(n_estimators=100, random_state=42)
- model.fit(X_train, y_train)
-
- # predict dependent variable
- y_pred = model.predict(X_test)
-
- # r²-value calculation
- r2 = r2_score(y_test, y_pred)
- print("r² Score:", r2)
-
- # Feature Importance
- importance = model.feature_importances_
-
- # sort features according to importance
- s_id = np.argsort(importance)
- pos = np.arange(s_id.shape[0])
-
- # MinMax scaling
- scaler = MinMaxScaler()
- importance_scaled = scaler.fit_transform(importance.reshape(-1, 1)).flatten()
-
- # importance
- total_importance = np.sum(importance_scaled)
- percentage_importance = (importance_scaled / total_importance) * 100
-
- # show top ten
- top_n = 10
- s_id = s_id[-top_n:]
- features = pd.DataFrame(
- {'index1': np.array(X_train.columns)[s_id], 'importance_normalized': importance_scaled[s_id],
- 'percentage_importance': percentage_importance[s_id]})
- return features
-
-
- def plot_feature_importance(self, features, path: str, base_width: int =10, base_height: int =6):
- """\
- Plots the ten most important features and returns a pyplot.
- Needs the ten top features and their importances as parameters.
-
- Parameters
- ----------
- features
- DataFrame returned by function feature_importance
- path
- path to dict where plot should be saved
- base_width
- width of plot
- base_height
- height of plot
-
- Returns
- ----------
- Show plot
-
- """
-
- num_features = len(features)
-
- plot_width = base_width
- plot_height = base_height + 0.2 * num_features
-
- ax = features.plot.bar(x='index1',
- y='importance_normalized',
- color='darkgray',
- legend=False,
- figsize=(plot_width, plot_height),
- width=0.8, fontsize=20)
-
- plt.xlabel('')
- plt.ylabel('Importance', fontsize=20)
-
- #adjust text height
- for i, v in enumerate(features['percentage_importance']):
- if features['importance_normalized'][i] + 0.01 > 1.1:
- text_height = 1
- else:
- text_height = features['importance_normalized'][i] + 0.01
- ax.text(i, text_height, f'{v:.1f}%', ha='center', va='bottom', fontsize=16, color='black')
-
- plt.title('Top 10 Features', fontsize=30, loc='left')
- plt.ylim(0, 1.1)
- plt.xticks(rotation=45, ha='right')
-
- if path is not None:
- try:
- plt.savefig(path, dpi=300, bbox_inches='tight')
- print(f"Plot successfully saved to {path}")
- except Exception as e:
- print(f"An error occurred while saving to png: {e}")
-
- return plt.show()
-
- def cluster_kmeans(self, n_cluster: int, label_x: str, label_y: str):
- """\
- Cluster self.df by kmeans clustering and show result as plt.show().
-
- Parameters
- ----------
- number_cluster
- number of clusters
- label_x
- x-axis label
- label_y
- y-axis label
-
- Returns
- ----------
- Show plot
- """
-
- kmeans_labels = cluster.KMeans(n_clusters=n_cluster).fit_predict(self.df)
-
- #plot
- plt.style.use('seaborn-v0_8-poster')
- plt.figure(figsize=(6, 6))
-
- plt.scatter(self.df[['x']],
- self.df[['y']],
- c=kmeans_labels,
- s=1,
- cmap='Set1');
-
- plt.title('K-Means Clustering')
- plt.xlabel(label_x)
- plt.ylabel(label_y)
- plt.xticks([])
- plt.yticks([])
-
- self.df['kmeans_cluster'] = kmeans_labels.tolist()
- return plt.show()
-
- def cluster_gmm(self, number_component: int, random_s: int, label_x: str, label_y: str):
- """\
- Cluster self.df by Gaussian Mixture Modeles and plot result.
-
- Parameters
- ----------
- number_component
- number of components of gmm clustering
- random_s
- random state of gmm clustering
- label_x
- x-axis label
- label_y
- y-axis label
-
- Returns
- ----------
- Show plot
-
- """
-
- gmm = GaussianMixture(n_components=number_component, random_state=random_s)
-
- gmm.fit(self.df)
- gaussian_labels = gmm.predict(self.df)
-
- #plot
- plt.style.use('seaborn-v0_8-poster')
- plt.figure(figsize=(6, 6))
-
- plt.scatter(self.df[['x']],
- self.df[['y']],
- c=gaussian_labels,
- s=1,
- cmap='Set1');
-
- plt.title('Gaussian Mixture Clustering')
- plt.xlabel(label_x)
- plt.ylabel(label_y)
- plt.xticks([])
- plt.yticks([])
-
- self.df['GMM_cluster'] = gaussian_labels.tolist()
- return plt.show()
-
- def cluster_hdbscan(self, cluster_size: int, label_x: str, label_y: str):
- """\
- Cluster self.df by hdbscan and plot result.
-
- Parameters
- ----------
- cluster_size
- size of clusters
- label_x
- x-axis label
- label_y
- y-axis label
-
- Returns
- ----------
- Show plot
-
- """
-
- clusterer = hdbscan.HDBSCAN(min_cluster_size=cluster_size, gen_min_span_tree=True)
-
- clusterer.fit(self.df)
- hdbscan_labels = clusterer.labels_
-
- # Plot
- outliers_mask = hdbscan_labels == -1
- plt.style.use('seaborn-v0_8-poster')
- plt.figure(figsize=(6, 6))
-
- plt.scatter(self.df[['x']],
- self.df[['y']],
- c=hdbscan_labels,
- cmap='Spectral',
- s=5)
-
- plt.scatter(self.df.loc[outliers_mask, 'x'],
- self.df.loc[outliers_mask, 'y'],
- s=4,
- c='gray',
- marker='v',
- label='Outliers',
- alpha=0.5)
-
- plt.title('HDBSCAN Clustering')
- plt.xlabel(label_x)
- plt.ylabel(label_y)
- plt.legend(markerscale=6)
-
- plt.xticks([])
- plt.yticks([])
-
- self.df['hdbscan_cluster'] = hdbscan_labels.tolist()
- return plt.show()
-
-
-
- #def for dmap/umap plots
-
- def check_dataframe(self):
- """\
- check if df is empty
- """
- if self.df.empty:
- raise ValueError("Dataframe is empty.")
-
- def prepare_data_source(self):
- """\
- set ColumnDataSource as self.df and return ColumnDataSource
- """
- return ColumnDataSource(self.df)
-
- def configure_hover_tooltips(self, feature: str, hover_tooltips: list[tuple[str, str]] = None):
- """\
- Set hover tooltips. Either standard or personalized.
-
- Parameters
- ----------
- feature
- column of dataset
- hover_tooltips
-
- Returns
- ----------
- HoverTool
-
- """
- if hover_tooltips is None:
- hover_tooltips = [
- ("Feature", f"{feature}"),
- ("Index", "ID"),
- ("X-Value", "x"),
- ("Y-Value", "y")
- ]
-
- hover_tooltips_formatted = "".join([
- f"{label}: @{field}
"
- for label, field in hover_tooltips
- ])
-
- hovertool_kwargs = {
- "name": "data",
- "tooltips": hover_tooltips_formatted
- }
- return HoverTool(**hovertool_kwargs)
-
- def create_base_plot(self, fig_width: int,
- fig_height: int,
- fig_title: str,
- label_x: str,
- label_y: str,
- range_x: list[float],
- range_y: list[float],
- tools_emb,
- title_align: str):
- """\
- Settings for plotting umap/densmap.
-
- Parameters
- ----------
- fig_width
- width of figure
- fig_height
- height of figure
- fig_title
- title of figure
- label_x
- x-axis label
- label_y
- y-axis label
- range_x
- range of x-axis
- range_y
- range of y-axis
- tools_emb
- tools that should be embedded
- title_align
- position of title
-
- Returns
- ----------
- plot
-
- """
- if not bokeh_above_three:
- plot_kwargs = dict(
- plot_width=fig_width,
- plot_height=fig_height,
- title=fig_title,
- tools=tools_emb,
- x_axis_label=label_x,
- y_axis_label=label_y,
- x_range=Range1d(start=range_x[0], end=range_x[1]),
- y_range=Range1d(start=range_y[0], end=range_y[1])
- )
- else:
- plot_kwargs = dict(
- width=fig_width,
- height=fig_height,
- title=fig_title,
- tools=tools_emb,
- x_axis_label=label_x,
- y_axis_label=label_y,
- x_range=Range1d(start=range_x[0], end=range_x[1]),
- y_range=Range1d(start=range_y[0], end=range_y[1])
- )
-
- plot = figure(**plot_kwargs)
-
- # configure title
- plot.title.align = title_align
- plot.title.text_font_size = '30pt'
-
- return plot
-
- def configure_axes_and_legend(self, plot, show_axes: bool, show_legend: bool):
- """\
- Settings for axes and legend.
-
- Parameters
- ----------
- plot
- show_axes
- add axes
- show_legend
- add legend
-
- Returns
- ----------
- None
-
- """
-
- #axes
- if not show_axes:
- plot.xaxis.visible = False
- plot.yaxis.visible = False
-
- plot.grid.visible = False
- plot.outline_line_color = None
-
- plot.xaxis.axis_label_text_font_size = "20pt"
- plot.yaxis.axis_label_text_font_size = "20pt"
-
- plot.xaxis.ticker = []
- plot.yaxis.ticker = []
-
- # legend
- if show_legend:
- plot.legend.title_text_font_style = "bold"
- plot.legend.background_fill_alpha = 0.0
- plot.legend.border_line_alpha = 0
- plot.legend.label_text_font_size = '20pt'
- plot.legend.title_text_font_size = '20pt'
- plot.legend.glyph_height = 30
- plot.legend.glyph_width = 30
- plot.add_layout(plot.legend[0], 'center')
- else:
- plot.legend.visible = False
-
-
- def cat_plot(self, feature: str,
- subs: list[str],
- colors: list[str],
- outputf: str,
- fig_width: int,
- fig_height: int,
- fig_title: str,
- label_x: str,
- label_y: str,
- range_x: list[float],
- range_y: list[float],
- hover_tooltips: list[tuple[str, str]] = None,
- show_legend: bool = False,
- point_size: int = 10,
- point_alpha: float = 0.6,
- show_axes: bool = False,
- title_align: str = 'center') -> None:
-
- """\
- Create plot with categorical color mapper.Choose feature, colors and more.
- Loads html file.
-
- Parameters
- ----------
- feature
- feature for categorical mapper
- subs
- values of feature
- colors
- list of colors
- outputf
- path to outputfile
- fig_width
- width of figure
- fig_height
- height of figure
- fig_title
- title of figure
- label_x
- x-axis label
- label_y
- y-axis label
- range_x
- x-axis range
- range_y
- y-axis range
- hover_tooltips
- hover tooltips
- show_legend
- possibility to add legend
- point_size
- size of points
- point_alpha
- alpha of points
- show_axes
- add axes
- title_align
- position of title
-
- Returns
- ----------
- show plot
-
- """
-
- self.check_dataframe()
- output_file(outputf)
-
- source = self.prepare_data_source()
- hover_emb = self.configure_hover_tooltips(feature, hover_tooltips)
- cm = CategoricalColorMapper(palette=colors, factors=subs)
-
- tools_emb = ['save', 'lasso_select', 'pan', 'wheel_zoom', 'reset', hover_emb]
- plot = self.create_base_plot(fig_width,
- fig_height,
- fig_title,
- label_x,
- label_y,
- range_x,
- range_y,
- tools_emb,
- title_align)
-
- plot.circle('x', 'y',
- size=point_size,
- color={'field': feature, 'transform': cm},
- alpha=point_alpha,
- source=source,
- name="data",
- legend_group=feature)
-
- self.configure_axes_and_legend(plot, show_axes, show_legend)
-
- show(plot)
-
- def lin_plot(self,
- outputf: str,
- feature: str,
- colors: list[str],
- fig_width: int,
- fig_height: int,
- fig_title: str,
- label_x: str,
- label_y: str,
- range_x: list[float],
- range_y: list[float],
- hover_tooltips: list[tuple[str, str]] = None,
- show_legend: bool = False,
- point_size: int = 3,
- point_alpha: float = 0.7,
- show_axes: bool = True,
- title_align: str = 'center') -> None:
-
-
- """\
- Create plot with Linear color mapper.Choose feature, colors and more.
- Loads html file.
-
- Parameters
- ----------
- outputf
- path to output file
- feature
- feature for linear mapper
- colors
- list of colors
- fig_width
- width of figure
- fig_height
- height of figure
- fig_title
- title of figure
- label_x
- x-axis label
- label_y
- y-axis label
- range_x
- range of x-axis
- range_y
- range of y-axis
- hover_tooltips
- hover tooltips
- show_legend
- add legend
- point_size
- size of points
- point_alpha
- alpha of points
- show_axes
- add axes
- title_align
- title position
-
- Returns
- ----------
- show plot
-
- """
-
-
- self.check_dataframe()
- output_file(outputf)
-
- source = self.prepare_data_source()
- hover_emb = self.configure_hover_tooltips(feature, hover_tooltips)
- lm = LinearColorMapper(palette=colors, low=min(self.df[feature]), high=max(self.df[feature]))
-
- tools_emb = ['save', 'lasso_select', 'pan', 'wheel_zoom', 'reset', hover_emb]
- plot = self.create_base_plot(fig_width,
- fig_height,
- fig_title,
- label_x,
- label_y,
- range_x,
- range_y,
- tools_emb,
- title_align)
-
- plot.circle('x', 'y',
- size=point_size,
- fill_color={'field': feature, 'transform': lm},
- alpha=point_alpha,
- line_alpha=0,
- line_width=0.03,
- source=source,
- name="data",
- legend_group=feature)
-
- self.configure_axes_and_legend(plot, show_axes, show_legend)
-
- show(plot)
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-