4.analyze_data

4. Analysis of CFReT Data

In this module, we perform analysis of the CFReT data to reach our goals as specified in the main README.

In the notebooks folder, we have two different folders for analysis:

• linear_model: Perform linear modeling per CellProfiler feature to determine which features significant depending on the co-variates used.

• UMAP: Generate UMAPs labeling different metadata per plate to assess if there is any clustering of morphology features.

• histogram_plot: Generate histogram plot comparing the number of neighbors adjacent to each single-cell per heart number to view the distribution of neighbors across hearts.

Linear Model (LM)

In this folder, we have four different types of notebooks labeled by number:

#0

In the 0 notebooks, we fit the linear model to each CellProfiler feature either to all the data or stratifying the data to only include cells with specific metadata values.

#1

In the 1 notebooks, we perform two different tasks. One task is to find single-cell crops of the top 5 values in a given CellProfiler feature based on the LM results. The other is to visualize the LM beta coefficients as a scatter plot.

#2

In the 2 notebook, we perform a power analysis to determined if we are fully powered given the number of single-cells that were segmented per plate.

#3

In the 3 notebook, we visualize the power analysis as a scatter plot.

UMAP

In this folder, we have three different types of notebooks labeled by number:

#0

In the 0 notebook, we use the UMAP method to reduce the morphology space and output a CSV file per plate.

#1

In the 1 notebooks, we visualize the UMAP coordinates for either all plates, only plate 3, or only plate 4. We label the points with different metadata based on the plate.

#2

In the 2 notebooks, we generate random single-cell crops per cluster of the UMAP, in which the UMAP data frame is stratified based on the range of coordinates of the cluster. One notebook is meant for plate 3 data and the other is for plate 4 data.

Histogram plot

In this folder, there is only one notebook that is used to load in the annotated data frame from plate 4, and then create a histogram plot with all heart number distributions on it to compare.

Create environments

In this module, there are two different environments to create to use in the notebooks.

• python_analysis_env: This environment is for use in python specific notebooks.
• R_analysis_env: This environment is for use in R specific notebooks.

You can create the environments using the code block below:

# create environment for analysis
mamba env create -f python_analysis_env.yml