Filtering reads by quality score

451 views
Skip to first unread message

John Blischak

unread,
Apr 8, 2015, 10:17:45 AM4/8/15
to rna-...@googlegroups.com
Hi,

I'd like to have STAR only output the uniquely mapping reads. I thought I could accomplish this using the option --outFilterScoreMin. It's documentation is below.

--outFilterScoreMin
  default: 0
  int: alignment will be output only if its score is higher than this value

If it is supposed to filter the reads, I am doing something wrong. If I pass the integer 255 (or 254), I get no output. If I pass the integer 10, I get no filtering (i.e. even reads with score less than 10 are output).

What should I pass as an argument to STAR such that only reads with mapping score 255 (i.e. mapped uniquely) are output?

Please let me know if my mistake is not obvious, or if my goal is unclear, and I'll post a small example.

Thanks,

John

Alexander Dobin

unread,
Apr 8, 2015, 11:26:33 PM4/8/15
to rna-...@googlegroups.com
Hi John,

--outFilterScoreMin filters reads by "alignment score", which is different from SAM's "mapping quality score". Alignment score is a sum of +/-1 for matches/mismatches and penalties for gaps, and it's recorded in the AS attribute.
To output only unique mappers you need to use --outFilterMultimapNmax 1

Cheers
Alex
Reply all
Reply to author
Forward
0 new messages