is there any way to normalize the counts and compare it for the splicing quantification?
1
0
Entering edit mode
9 days ago
PK ▴ 130

Hi All, I conducted paired-end total RNA sequencing with triplicates for both wild type and drug-treated conditions. Each wild type BAM file contains ~200 million mapped reads (processed with STAR using two-pass mode, read length 65 nt), while the treatment BAM files have ~400 million mapped reads. I analyzed splicing events, focusing on exon skipping, using rMATS-turbo (comparison-1).

To assess reproducibility, I randomly down sampled the reads in each BAM file to 90%, 80%, and 50% and repeated the splicing analysis (comparison-2). I compared the skipped exons identified in comparison-1 and comparison-2 but still observed many skipped exons.

Then i asked to rMATS developers how the splicing quantification was done between the sample group. The answer was it considers the junction counts but no normalization was done before comparison.

So, I'm asking is there any way to normalize the counts and compare it for the splicing quantification.
Also, i'm only considering the junction counts.

Thanks

majiq splicing rmats • 279 views
ADD COMMENT
2
Entering edit mode
9 days ago

I do not think that the normalization is necessary in your case.

I'm not sure what you asked the developers but given that the PSI value is a ratio it should be agnostic to library depth (at least by itself).

The confidence will decrease of course if you have very few reads in a given region after downsampling (which is part of the rMATS model). But on the backend you can already filter based on your own FDR threshold.

So you probably should just use a more conservative p-value/FDR cut-off if you believe your high depth is pulling out substantial artifacts.

See: https://groups.google.com/g/rmats-user-group/c/hpr7j9FMgFg?pli=1

ADD COMMENT
0
Entering edit mode

Very nice. I agree with this.

ADD REPLY

Login before adding your answer.

Traffic: 2532 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