SEQanswers

SEQanswers (http://seqanswers.com/forums/index.php)
-   Bioinformatics (http://seqanswers.com/forums/forumdisplay.php?f=18)
-   -   unmapped reads assigned to chromosomes (http://seqanswers.com/forums/showthread.php?t=15942)

efoss 12-01-2011 10:54 AM

unmapped reads assigned to chromosomes
 
If I use samtools' "idxstats" commands, it returns information with:

1. reference sequence name (chromosome name in my case)

2. length of reference sequence

3. number of mapped reads assigned to that reference sequence

4. number of unmapped reads assigned to that reference sequence

Number 4 makes no sense to me. If a read is unmapped then by definition you can't assign it to a genomic location. If I remove everything with map quality less than 1, most of these go away but not all of them, i.e. I still end up with unmapped reads assigned to chromosomes. Does anyone understand this?

Thank you.

Eric

swbarnes2 12-01-2011 11:08 AM

Come on. If you aren't going to read the SAM format specs before you ask a question about the SAM format, at least search the message board before asking a question. I think I've answered this question three times this week alone.

efoss 12-01-2011 11:16 AM

Quote:

Originally Posted by swbarnes2 (Post 58464)
Come on. If you aren't going to read the SAM format specs before you ask a question about the SAM format, at least search the message board before asking a question. I think I've answered this question three times this week alone.

Sorry. I had searched the message board unsuccessfully, but yes - it was stupid not to look at the sam documentation more closely:

4.1 For a unmapped paired-end or mate-pair read whose mate is mapped, the unmapped read should have RNAME and POS identical to its mate.

Eric

swbarnes2 12-01-2011 01:22 PM

Yup. The other possibility is if you use bwa for alignments. bwa concatenates all the reference sequences together, so if a read hangs off of one reference onto another, it will be given the right chromosome and position, but the 4 flag will also be put in, as a sign that something is not quite right. I'm not sure if other aligners, like bowtie, will do that.

efoss 12-01-2011 01:24 PM

Quote:

Originally Posted by swbarnes2 (Post 58485)
Yup. The other possibility is if you use bwa for alignments. bwa concatenates all the reference sequences together, so if a read hangs off of one reference onto another, it will be given the right chromosome and position, but the 4 flag will also be put in, as a sign that something is not quite right. I'm not sure if other aligners, like bowtie, will do that.

Thanks very much. That's good to know, since I am aligning with bwa.

Eric


All times are GMT -8. The time now is 12:17 AM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2019, vBulletin Solutions, Inc.