Different regions for normalisation
1
0
Entering edit mode
msrch • 0
@c3c20488
Last seen 18 months ago
Germany

Hello,

I am computing a differential analysis on count data per region with DESeq2. For example, comparing only the region close to the transcription initiation site. Therefore, my count table represents the counts for each gene that map within a given interval (e.g. [TSS, TSS + 300]) for each sample.

I wonder if it is correct to compute the size factors with counts in the complete interval ([TSS, pA]) and then using them for the differential analysis limited in the region mentioned above. I would provide the size factors rather than calling estimateSizeFactors() with the data in my interval.

Thanks in advance for any insights

DESeq2 median-of-ratios normalisation • 670 views
ADD COMMENT
1
Entering edit mode
ATpoint ★ 4.5k
@atpoint-13662
Last seen 27 minutes ago
Germany

Essentially, it does not matter how you get the size factors as long as they're accurate and it's reproducible. A good diagnostic is MA-plots on the output of results(). They typically have an arrowhead-like shape and the tip if the arrowhead should be centered at y=0. Overall, the majority of points should be centered there as well. Checking control genes (if available) can help as well.

ADD COMMENT
0
Entering edit mode

Thank you very much. This solved the issue. The MA plot helped to point me to the right size factors. Thanks again

ADD REPLY

Login before adding your answer.

Traffic: 1049 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6