search

gao-lab / GLUE

Graph-linked unified embedding for single-cell multi-omics data integration
MIT License
371 stars 56 forks source link

Question about preprocessing ATAC-Seq #116

Open luglilab opened 6 months ago

luglilab commented 6 months ago

Hello,

I'm interested in using scglue to integrate my scRNAseq with scATACseq data, comprising four paired samples. These datasets are not multiome; they're only paired samples.

After processing with Cellranger, I used episcanpy's epi.ct.bld_mtx_fly to construct a count matrix from the TSV/TBI files. Subsequently, I filtered the data using epi.pp.filter_cells and epi.pp.filter_features.

My current concern pertains to the input matrix for the scglue.data.lsi function.

Could you please advise on whether I should use the Raw Matrix, Binarized Matrix, or normalizedMatrix for scglue.data.lsi?

Thank you. Your assistance is appreciated.

Jeff1995 commented 6 months ago

Hi @luglilab. Thanks for your interest in GLUE!

I would recommend using the raw count matrix, as was also recently confirmed in this paper.

luglilab commented 6 months ago

Thanks a lot!

luglilab commented 5 months ago

Hello @Jeff1995 ,

could you suggest a good preprocessing pipeline for 10X data compatible with your tool? I'm trying to figure it out if is better snapatac, episcanpy or signac but i'd like to have your comment.

Thank you.

Jeff1995 commented 5 months ago

Hi @luglilab, I haven't really fiddled with different preprocessing pipelines. As long as it outputs count matrices, it would work well with scglue. scglue has builtin lsi implementation and does not need further processing than the count matrices.