Failed chastity reads

[blindtiger454]

Has anyone done RNA-Seq with reads that failed chastity test? When we sequence, we get a export file and fastq file back from Illumina. The export file usually has a few million reads extra that failed chastity and never made it into the fastq file. In one of our samples, 75% of the reads failed chastity. I assume it is safe to use these reads during mapping? If they are truly bad reads, they shouldn't map. Or does anyone disagree???

[kmcarr]

Has anyone done RNA-Seq with reads that failed chastity test? When we sequence, we get a export file and fastq file back from Illumina. The export file usually has a few million reads extra that failed chastity and never made it into the fastq file. In one of our samples, 75% of the reads failed chastity. I assume it is safe to use these reads during mapping? If they are truly bad reads, they shouldn't map. Or does anyone disagree???

First things first, if you I had 75% failed reads I would call that a failed run and request a re-run from your service provider.

Can you use the failed reads? Sure, you can but is it wise to do so. What is your experiment? If you are strictly doing expression analysis, that is counting fragments aligned to transcription units then you could probably get away with it. On the other hand if you are interested in identifying variations (e.g. SNPs) then I would not use failed reads. They would produce too many false positives. In fact if you are using your sequence data for variant discovery you should go beyond the Illumina chastity filters, cleaning up the reads even further.