Skip to content
Predicting Cell Health with Morphological Profiles
Jupyter Notebook HTML Other
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information. Adding Download Module (#71) Oct 18, 2019
1.generate-profiles rerun consensus generation Nov 14, 2019
2.replicate-reproducibility Adding Analysis Pipeline (#73) Oct 18, 2019
3.train add coefficient summary Nov 20, 2019
4.apply reduce file size of umap consensus pdf Nov 20, 2019
.gitattributes Adding Analysis Pipeline (#73) Oct 18, 2019
.gitignore Adding Platemap Visualization (#10) May 20, 2019
.travis.yml Adding tests and travis (#31) Aug 1, 2019
LICENSE fix readme figure link Oct 27, 2019
environment.yml update dplyr to 0.8 Nov 14, 2019

Predicting Cell Health with Morphological Profiles

Gregory Way, Shantanu Singh, Anne Carpenter 2019


The following repository stores a complete analysis pipeline of using Cell Painting data to predict the results of several cell health assays.

Our collaborators previously collected a series of assays measuring different aspects of cell health. The assays include staining with specific dyes to measure specific phenotypes.

Assay/Dye Phenotype
Caspase 3/7 Apoptosis
DRAQ7 Cell Death
EdU Cellular Profileration
Hoechst DNA Content
Ph3 Cell Division
gH2Ax DNA Damage

We hypothesized that we can use unbiased and high dimensional Cell Painting profiles to predict the readouts of each individual assay.


We collected Cell Painting measurements on a CRISPR experiment. The experiment targeted 59 genes, which included 119 unique guides (~2 per gene), across 3 cell lines. The cell lines included A549, ES2, and HCC44.

CRISPR Correlation

About 40% of all CRISPR guides were reproducible. This is ok since we are not actually interested in the CRISPR treatment specifically, but instead, just its corresponding readout in each cell health assay.


We performed the following approach:

  1. Split data into 85% training and 15% test sets.
  2. Normalized data by plate (z-score).
  3. Selected optimal hyperparamters using 5-fold cross-validation
  4. Trained elastic net regression models to predict each of the 70 cell health assay readouts, independently.
  5. Trained using shuffled data as well.
  6. Report performance on training and test sets.

We also trained logistic regression classifiers using the same approach above.


Regression Model Performance

Initial results indicate that many of the cell health phenotypes can be predicted with our approach.

Computational Environment

We use conda as a package manager. To install conda see instructions.

To create the computational environment, run the following:

# Make sure the repo is cloned
conda env create --force --file environment.yml
conda activate cell-health

Analysis Modules

Each analysis module should be run in order. View each module for specific instructions on how to reproduce results.

You can’t perform that action at this time.