Hi,
I am currently testing a development bioinformatics pipeline that uses BWA-MEM aligner.
For some reason, I was under the impression that the MAPQ (score for single end read) from BWA-MEM is capped at 60.
However, in a few samples I'm seeing a MAPQ of 70 in regions that show MAPQ = 60 in the rest of the samples.
Is this a potential bug in this development pipeline? Or does BWA-MEM really produce MAPQ values of 70? And is there literature available on this?
Thanks
•
written
by
Saha.Neiladri •
Don't read too much into the MAPQ scores. These do not represent accurate measures and are more of "guesstimates."
Aligner developers apply certain heuristics using data estimates how often the alignment is correct, but the actual alignment correctness depends on the sequence and some sequences will never be reported incorrectly, hence the entire probability model is wrong.
A few MAPQ scores may have special meaning for example MAPQ=0 means multiply mapped read.
Note how the MAPQ=0 is actually wrong in that case, the chance of reporting the alignment incorrectly is NOT actually zero but depends on how many other equal positions are there.
Indeed, MAPQ is a 'messy case', whereby it is measured differently in different programs. I think that, in the official SAM specification, it is measured in the range 0 - 255.
Log in to answer this question.
