softclip vs mismatch at either end

Question

[-_-]

I noticed that in my BWA alignments, given a read that is mostly well aligned but with a few bases at the end mismatched, sometimes it's labeled as a mismatch (M), while at other time it's a softclipping (S). What's the deciding factor for mismatch vs soft clipping?

For example, I have seen cases where two 76bp reads are fully aligned except for the first base, yet in one case the first base is a called as a mismatch, but in the second case it's soft clipped. I wonder what are the possible reasons for such difference.

[michaela_boell]

since you posted this three months ago: did you come up with some theories/sources?

[-_-]

No, I still have no answer.

[michaela_boell]

I guess, then there is only the "make test files and see what happens" left.