Re: abyss-map SAM MAPQ?

18 views
Skip to first unread message

Dan Browne

unread,
Feb 7, 2016, 3:52:43 PM2/7/16
to Shaun Jackman, ABySS
Oh wait, I just re-read your email and think I figured it out. Anything with a score of 0 is a multimap, so basically anything with an alignment score above the minimum threshold (-l) is mapped uniquely?

Dan

On Sun, Feb 7, 2016 at 12:52 PM, Dan Browne <dbrow...@gmail.com> wrote:
Hey Shaun,

Thanks for the information. I've copied the ABySS Users Google Group on this message, for posterity. Follow up question: how does abyss-map identify reads as uniquely mapped? Is this property indicated in the SAM FLAG column?

Best,
Dan

On Fri, Feb 5, 2016 at 5:19 PM, Shaun Jackman <sjac...@gmail.com> wrote:

Hi, Dan. I don’t believe the MAPQ values of abyss-map are documented anywhere. Sorry for that. ABySS does not calculate a proper MAPQ value. To do that it would need to keep track of the second-best hit, and it doesn’t do that. The MAPQ value of abyss-map is the same as the alignment score, which is the number of matched nucleotides, or 0 in the case of a multimap. The MAPQ value is capped at 254.

I’m happy to answer e-mail questions, but if you send your question instead to BioStars, GitHub or Google Groups, the question and answer will be publicly archived for future Googlers.

Cheers,
Shaun




On 2016-February-05 at 13:08:10 , Dan Browne (dbrow...@gmail.com) wrote:

Hi Shaun,

I've been experimenting with abyss-map recently, capturing the standard SAM output as follows (ABySS version 1.9.0):

abyss-map -v -j20 -l125 ../../02_Separated_Pairs/Library.SXPX.L.fq.gz ../../02_Separated_Pairs/Library.SXPX.R.fq.gz ../../SXPX_SEQ1K_99-3.fa | abyss-fixmate --all --qname -v | samtools view -@20 -Sb - | samtools sort -@ 20 -O bam -T MAP_SXPX_SEQ1K_99_OLC > MAP_SXPX_SEQ1K_99_OLC.bam && samtools index MAP_SXPX_SEQ1K_99_OLC.bam

See the attached file for the inspection of the SAM file and the resulting MAPQs. I've never seen such high MAPQ values, and I am just wondering how abyss-map calculates the MAPQ? Am I missing something trivial?

I look forward to your insight.

Best,
Dan
 



Dan Browne

unread,
Feb 7, 2016, 3:52:43 PM2/7/16
to Shaun Jackman, ABySS

Shaun Jackman

unread,
Feb 7, 2016, 4:01:00 PM2/7/16
to ABySS, sjac...@gmail.com
Hi, Dan.

Correct, anything with a score of 0 is multimapped. Anything with an alignment score above 0 is mapped uniquely. There will be no alignments with a score between 1 and the minimum alignment score threshold specified by -l.

Cheers,
Shaun

On Sunday, 7 February 2016 12:52:43 UTC-8, Dan Browne wrote:
Oh wait, I just re-read your email and think I figured it out. Anything with a score of 0 is a multimap, so basically anything with an alignment score above the minimum threshold (-l) is mapped uniquely?

Dan
On Sun, Feb 7, 2016 at 12:52 PM, Dan Browne wrote:
Hey Shaun,

Thanks for the information. I've copied the ABySS Users Google Group on this message, for posterity. Follow up question: how does abyss-map identify reads as uniquely mapped? Is this property indicated in the SAM FLAG column?

Best,
Dan
On Fri, Feb 5, 2016 at 5:19 PM, Shaun Jackman wrote:

Hi, Dan. I don’t believe the MAPQ values of abyss-map are documented anywhere. Sorry for that. ABySS does not calculate a proper MAPQ value. To do that it would need to keep track of the second-best hit, and it doesn’t do that. The MAPQ value of abyss-map is the same as the alignment score, which is the number of matched nucleotides, or 0 in the case of a multimap. The MAPQ value is capped at 254.

I’m happy to answer e-mail questions, but if you send your question instead to BioStars, GitHub or Google Groups, the question and answer will be publicly archived for future Googlers.

Cheers,
Shaun

Reply all
Reply to author
Forward
0 new messages