This repo hosts critical scripts associated with the paper titled, CaSSiDI: novel single-cell "Cluster Similarity Scoring and Distinction Index" reveals critical functions for PirB and context-dependent Cebpb repression https://www.nature.com/articles/s41418-024-01268-8
Currently, we provide three scripts that contain the following:
-
Step1_Seurat_processing_steps.R
lists the steps we followed to process the 10X scRNAseq data from the raw matrices to clusters and marker genes. -
Step2_CaSSiDI_Scores_1_to_7.R
is a function to compute the cluster similarity scores (CSS) 1 to 7, described in the paper. These are the simpler (intermediate) versions of the final CSS 8, that was finally chosen as the best performing score. -
Step2_CaSSiDI_Score8.R
is a function to compute Score 8, which was the final chosen score.
We are currently building the next version of these scripts with the goal of providing a complete automatic pipeline in a Linux Slurm environment using Bash scripting. This pipeline can be used to run parallel Seurat jobs for each combination of nPCs and clustering resolution parameter so that a complete grid-search style optimization can be carried out. The final steps of the pipeline would include post-processing scripts to compute the CSS scores and distinction index for any pair of scRNAseq datasets.
Please check back soon!