aggregating peaks without 10x feature linkage

[jhong26]

Hello, I am analyzing 10x multiome data and looking to aggregate macs2 called peaks so I cannot use the 10x-generated feature_linkage.bedpe file to aggregate peaks. As suggested in a previous post, I plan to use cicero to link peaks to other peaks. However, the 10x feature linkage file also identifies peak-gene links combining with gene expression data from scRNAseq. I wonder if I should combine peak-peak links from cicero with gene-peak links from Signac (LinkPeaks) or Archr (addPeak2GeneLinks) to feed into the MultiVelo's aggregate peaks function. Or should I just use the peak-peak links from cicero? Thank you, Jason

[danielee0707]

Hi. All the above approaches are valid solutions. The second approach can boost the ATAC signal a little bit more. You can create a custom feature_linkage file combining the peak-peak and peak-gene or gene-peak links and run the aggregate peaks function. Because I have actually never used these functions myself, it may be worth checking the quality and confidence of the gene-peak links and then manually setting a threshold for cutoff.

[jhong26]

Thanks so much!

On Thu, Aug 10, 2023, 2:09 PM Chen Li @.***> wrote:

Hi. All the above approaches are valid solutions. The second approach can boost the ATAC signal a little bit more. You can create a custom feature_linkage file combining the peak-peak and peak-gene or gene-peak links and run the aggregate peaks function. Because I have actually never used these functions myself, it may be worth checking the quality and confidence of the gene-peak links and then manually setting a threshold for cutoff.

— Reply to this email directly, view it on GitHub https://github.com/welch-lab/MultiVelo/issues/29#issuecomment-1673923315, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHVWKGZDEIINJ6UBRUGTS5LXUVEXFANCNFSM6AAAAAA3KVXUSM . You are receiving this because you authored the thread.Message ID: @.***>