I am using chimeric reads for my analyses, I am asking some questions in a row these days so I hope I'm not getting annoying :D
When bwa mem
generates a sam output, it can include supplementary alignments (distinguished by the flag 2048). If I sum up the alignment length of the main alignment and of the supplementary alignment of each read that has one, sometimes I find that it goes beyond the read length.
I would like bwa mem
to search for a supplementary alignment only with the unmapped portion of a read: is this possible?
This is the interesting part :)
bu if bwa was able to map this small unmapped fragment, it would be in the SA tag. Why do you expect a new result from BWA ?
Let's say this is the read:
We have perhaps one alignment here:
And another here:
Putting them together:
They don't overlap on the read sequence. However, in my dataset I find sometimes situations like this one:
Where you can see the overlap in the middle. This means that a region of the read is part of both the "first" alignment and the supplementary alignment. Obviously, I don't want this, because I want each position of the read to be mapped only once.