Illumina quality bias 5' end

[mapardo]

I have 50 bp Illumina RNA-seq reads that display a strange behaviour.
Per base sequence quality has not the usual trend for Illumina data.
The base with the highest is the number 14, and at the 5' end, quality raises in a step way.
Has anyone seen this behaviour?

[dwmohr]

While there could be other factors, my guess would be uneven base composition for your library. The first few cycles are used to estimate crosstalk and phasing, and the software expects an even representation of bases. If my guess is correct, you can mitigate this problem by running a control lane, or by reanalyzing the data from intensities and applying a previously run control lane.

Hope this helps,

David

[pmiguel]

I have 50 bp Illumina RNA-seq reads that display a strange behaviour.
Per base sequence quality has not the usual trend for Illumina data.
The base with the highest is the number 14, and at the 5' end, quality raises in a step way.
Has anyone seen this behaviour?

Yes, this is now normal for HiSeq/HiScanSQ data as of the software upgrade that allows use of the v3 sequencing chemistry. See:

http://seqanswers.com/forums/showthread.php?t=13155

For more discussion.

--
Phillip

[mapardo]

Thank you for such a quick answer.
Things are clearer for me now.

[dwmohr]

Interesting thread (I missed this previously).

Interestingly enough, we do NOT see this pattern for our exome/genome samples, only RNAseq or samples with known skewed base composition in the first few cycles. Attached is a plot of a recent exome sample as an example. We're running v3 chemistry with the latest HCS software.