This repository includes the main scripts for my master thesis and on-going project about single-cell morphological data.
Result dataframes and plots are not included in this repository!
Mechanism of action classification for two SPECS5K subsets. Main training script:
mlp_classification.py
Remaining scripts contain the evaluation of training results and statistical calculations.
Analysis scripts for unsupervised Beactica and SPECS5K analysis. Additionally includes normalization scripts, grit calculations, cell viewer as well as standard UMAP analysis.
Specific unsupervised analysis involving scanpy and cellxgene framework. Specifically, usage of PAGA embeddings and leiden/louvain clusterings. E distance and testing also included here.
Code base for single-cell bokeh application developped with Dan. Use
generate_single_cell_bokeh.py
for viewer .html with custom data sets.