Gene Regulatory Network in scverse

Dear all,
I was wondering if you have any pointers towards Gene Regulatory Network Analysis tools compatible with scverse (basically python based). Ideally, I would like to tap into both, single-cell and spatial transcriptomics (also combine it with GWAS summary stats, but thats maybe too much to ask).

If you have any tools or ideas that would be much appreciated.
The aim is to find gene networks co-regulated in a cell-type specific manner supported with Spatial Transcriptomics Data (and GWAS summary stats if available).

Lots to ask, I know …

1 Like

Hi @sim

Great question! May I ask if you also have scATAC-seq in your data? Depending on your answer there are different options:

You have scATAC-seq data: In this case, you can leverage methods that integrate both RNA-seq and ATAC-seq to infer GRNs. The ones available in python (by alphabetical order) are CellOracle (Welcome to CellOracle’s documentation! — celloracle 0.10.11 documentation), Dictys (GitHub - pinellolab/dictys: Context specific and dynamic gene regulatory network reconstruction and analysis) and SCENIC+ (Welcome to SCENIC+’s documentation! — SCENIC+ 0.1.dev132+g26f1691 documentation). Since all these methods build TF-Peak-Gene triplets, you would be able to enrich for GWAS terms that are present in regulatory peaks. Once you have your GRN (or cell-type GRNs) obtained from your single-cell data, you can use it to estimate TF activities in your spatial data using any statistical method. For example, SCENIC+ natively uses the method AUCell. In case you want to run other enrichment methods, the framework decoupler contains most of them, like classic GSEA or more modern ones: decoupler - Ensemble of methods to infer biological activities — decoupler 1.2.3 documentation. Then you can check how the obtained activities distribute across your tissue or how they relate to pathway activities, here is an example (Panel i) and a vignette showing how to do it: Spatial transcriptomics functional analysis — decoupler 1.2.3 documentation.

You don’t have scATAC-seq data: In case you only have scRNA-seq the options are more limited. You could still use CellOracle, since it allows to use non-specific prior peaks in the GRN inference process. Else you could use pySCENIC: GitHub - aertslab/pySCENIC: pySCENIC is a lightning-fast python implementation of the SCENIC pipeline (Single-Cell rEgulatory Network Inference and Clustering) which enables biologists to infer transcription factors, gene regulatory networks and cell types from single-cell RNA-seq data.. In this case the association to GWAS terms is way harder since you don’t have information about which genomic regions are relevant in gene regulation. Anyways, from the obtained GRNs you can still obtain TF activities and perform downstream analyses.

Hope this is helpful!