By mapping RNA seq data to a reference genome, then analyzed the mapped info using Picard, I got a median 5 prime bias of 0.6 and 3 prime of 0.3. The ratio is about 2. How do I understand these numbers? Good? Bad?
thanks,
wj
1 answer
As the description says, 5' (3') bias is the median of the following ratio:
[mean expression of 5' (3')] / [mean expression of whole transcript]
- Mean expression for 5' (3') is calculated as mean coverage of first (last) 100 bases
- Mean expression of transcript is the mean coverage of all bases in that transcript
- Median is calculated for the representative set of 1000 transcripts
This bias is typically introduced by library preparation (reverse transcription, etc). It is up to RNA-Seq analysis software to correct for it. For example see how Cufflinks handle this bias: http://cufflinks.cbcb.umd.edu/howitworks.html#hsbi
From this example below (Google images -> 5' to 3' RNA-seq bias:) it seems like you're in the "no bias low end coverage" situation. So if it is fine depends on what protocol have you used

Log in to answer this question.