How to prepare references for ATAC-only data (mm10)

[quiocarbajal]

Hello team!
Very nice paper!
I was trying to use your pipeline for scATAC data aligned to mm10 (using cellRanger, reference 2020-A, which is GENCODE vM23).
I am not sure on how to prepare data. In config file there are 3 reference files mentioned: 'gene_annotations', 'gene_TSS500', and 'genes'.
I am pretty sure that 'gene_TSS500' is necessary (and how to prepare it). What I am not sure is if 'gene_annotations' is necessary, and if 'genes' is necessary (or how should I prepare it, didn't find any reference to it).
Thank you very much!

[mayasheth]

Hi, thanks for your interest in the model!

We have created the reference files needed for this pipeline that are compatible with mm10 and uploaded them to the ABC codebase here.

You will need to modify the main scE2G config as well as config files in the submodules to make this work:

1. In the scE2G's config/config.yaml update the following:

• gene_annotations should point to the GENCODE vM23 GTF file
• gene_TSS500 can point to gencode.vM25.protein_coding.TSS500bp.bed from the above link
• genes can point to gencode.vM25.protein_coding.genes.bed

2. In the ENCODE_rE2G/config/config.yaml update the following (the other references will be overriden by the scE2G config, so you can leave them unchanged)

• gene_classes: gencode.vM25.protein_coding.gene_promoter_class.tsv (this file includes the "ubiquitously expressed gene" annotations we have for human genes mapped to mouse orthologs)

3. In ENCODE_rE2G/ABC/config/config.yaml, update the following (again, other references will be overridden by the higher submodules, so you can leave them unchanged)

• ref[chrom_sizes] should point to mm10.chrom.sizes
• ref[regions_blocklist' should point to mm10.blacklist.ENCFF547MET.bed

Apologies this is so complicated, we are working on streamlining the config files to make running in a different species more simple!

Please let us know if you have any issues.