Identifying Reproducible Transcription Regulator Coexpression Patterns with Single Cell Transcriptomics

https://doi.org/10.1371/journal.pcbi.1012962

This project looks to identify which gene partners are most commonly coexpressed with each TR in human and mouse, across a large corpus of single cell RNA-seq data. This information is compared to literature curated targets from low-throughput experiments, as well as aggregated ChIP-seq binding scores.

A main deliverable is the summarized/ranked information for each TR, which can be found at: https://borealisdata.ca/dataset.xhtml?persistentId=doi:10.5683/SP3/HJ1B24

NOTE: Analysis relies on objects (i.e., the underlying scRNA-seq datasets and resulting gene x gene coexpression matrices) that live on the Pavlab servers and are not easily shareable to a data repo given their size. Please contact me if you have any questions or requests!

Pavlab: The data and the associated download scripts are found:

/cosmos/data/downloaded-data/sc_datasets_w_supplementary_files/lab_projects_datasets/amorin_sc_datasets

Once scRNA-seq data was acquired, the scripts used to preprocess and build an aggregate coexpression network for each dataset are found in this repo:

R/preprocessing_scripts/CPM/

The Unibind ChIP-seq data used for binding evidence were downloaded and summarized using: https://github.com/PavlidisLab/Unibind_analysis/

For convenience, the Borealis repo above also contains the gene x experiment binding matrices used to generate the aggregate binding summaries.