search

GreenleafLab / chromVAR

chromatin Variability Across Regions (of the genome!)
https://greenleaflab.github.io/chromVAR/
Other
159 stars 38 forks source link

two samples for normal ATAC-Seq #33

Open dwb0211 opened 6 years ago

dwb0211 commented 6 years ago

Is it applicable for two samples to calculate the variability in normal ATAC-Seq but not single cell sequence?

AliciaSchep commented 6 years ago

I'm not sure I understand the question -- do you have two samples total?

dwb0211 commented 6 years ago

Yes, I have one sample for control from normal ATAC-Seq in cultured cells and one for treatment group. chromvar is finished with no error, I want to know whether this result is right or not. The figure is attache as below:

chromvar_atac

caleblareau commented 6 years ago

I would say that it's technically right in that the variability is a measure that can be applied to two samples. However, you probably almost certainly have some technical confounder there. How deeply did you sequence these two samples (or what's the colSums of the counts matrix)? Are the hyper-variable motifs matching at many or a few places?

dwb0211 commented 6 years ago

@caleblareau Thanks you so much for your explantation. There are 20M reads per sample and the result match with other methods.

ATpoint commented 5 years ago

I would say that it's technically right in that the variability is a measure that can be applied to two samples. However, you probably almost certainly have some technical confounder there. How deeply did you sequence these two samples (or what's the colSums of the counts matrix)? Are the hyper-variable motifs matching at many or a few places?

Could you give an example of such a technical confounding event? Is this based on you experience? We recently used chromVAR for a comparison between two bulk cell types (knockout vs wildtype situation) and the top-variable motifs (in both directions, = more and less accessable motifs) comprised both the knocked-out factor and TFs that were later supported by independent in silico and wetlab experiments.