Question

Can I calculate counts from the base mean and log2fc outputs of DESeq2? Also, can I calculate error for counts values from lfcse output?

0

Entering edit mode

robin.uchiyama • 0

@robinuchiyama-10844

Last seen 8.5 years ago

Hi Bioconductor community,

I performed a categorical analysis (control vs treated) using DESeq2, and I want to put together a figure where I show the average counts in the control and treated groups that includes error and a p value. To calculate counts of the treated group, do I simply multiply the log2 fc value (after first taking the anti-log2) by the base mean value (which is representative of control counts)? Do the standard error and adjusted p values only apply to the fc or can it be applied to counts?

Thanks!

deseq2 chipseq counts log2fc error • 2.2k views

ADD COMMENT • link updated 8.5 years ago by Michael Love 42k • written 8.5 years ago by robin.uchiyama • 0

1

Entering edit mode

There will be better replies I am sure, but I would just script that in R yourself i.e. calculation of mean counts for condition 1 versus 2 and standard error/ deviation whatever - it is a quick job even if you are a beginner. You just need to export the normalised counts from DESEQ then work from that. Then I am sure the p values apply to the counts because the fold change is calculated from the counts.

ADD REPLY • link 8.5 years ago chris86 ▴ 420

score 1 · Answer 1 · 2016-06-07

Take a look at the plotCounts function which is shown in the vignette. Rather than showing a summary (mean + SE) it's much preferred to just show the normalized counts themselves. If you have many samples (n > 20) you could also use a boxplot to summarize the normalized counts across groups. This is easily done with ggplot2, see the section of the vignette "Plot counts" that starts "For customized plotting,...".