More than one possible mapping place for a read

Question

[biotech]

When there is more than one possible mapping place for a read, being both places of equal score,

Where 'bwa mem' mapper finally maps it?

What about the mapping quality scores for the read?

Answer 1

[WouterDeCoster]

It depends on your settings, but I think the default is multimapping the reads. So the same read will be present in multiple locations. Mapping quality will be 0 (meaning the read maps equally well in multiple places).

[GenoMax]

Are you certain about multi-mapping? bwa mem manual entry does not say anything specific but the older sampe/samse explicitly say that

Repetitive read pairs will be placed randomly.

Repetitive hits will be randomly chosen.

[WouterDeCoster]

You are giving me doubts about my answer. ;-) Might very well be possible that random mapping happens.

Well, the easiest approach to settle this would be to test it. Have a single multimapping read, map it, count the reads in the samfile.

[biotech]

I really don't care about random mapping as long as having mapping quality of zero for them. In fact, multi-mapping reads should be discarded for some analysis such as RNA-seq.

[biotech]

Great proof test @decosterwouter, I will try to see what's going on!

[biotech]

I agree about that mapping quality is 0, but not sure if repetitive read pairs will be placed randomly.