Classify Single Cell Phenotypes From Images

CytoData Society

Image-based profiling can be used to detect mechanism of action (MOA) of small molecules. Here, we use single cell data from Breinig et al. 2015 to classify MOA for each single cell.

Traditionally, MOAs are predicted based on well-aggregated profiles. Here, we attempt to classify MOA of each individual single cell.

The following repository stores analytical code, data, computational environments, and pipelines to reproduce the full analysis.

Data

Pre-processed raw data is available at: https://doi.org/10.6084/m9.figshare.10247531.v1

Computational Environment

We use conda as an environment manager. Follow these instructions to download conda.

To initialize the compute environment used in this project, run:

# Using conda version > 4.7.12
# Step 1: Create the environment
conda env create --force --file environment.yml

# Step 2: Activate the environment
conda activate cytodata-single-cell

Analysis Pipeline

The analysis modules should be executed in order. Ensure that the cytodata-single-cell conda environment is activated.

See analysis.sh for more details.