Silly softclips in BWA

Question

[Aerval]

Hi all,

does anybody of you know why BWA sometimes produces softclips that have are an exact match of their aligned sequence

In the image above that shows a screen-shot from IGV you can see tumor reads aligned to a h37 reference genome (Sequence is shown on the bottom). Normal aligned sequences are shown in grey while softclips (and other mismatches like SNVs) are displayed by their respective bases. Whats unusual about the image above is that the softclip in the center matches the reference sequence and therefore should not be classified as softclip.?

I am analyzing structural variants with Socrates but get a lot of false positives through the described abnormality. This might be easy to filter out afterward but I would like to know whether there is a deeper reason for that.

Thanks for your help.

[Matt Shirley]

Can you illustrate where the soft clipping occurred? I don't see anything abnormal in your screenshot.

[Aerval]

I am not sure about that as these softclips seem absoultly random to me (but I have not checked the regions or anything similar). The only notable thing to me is that a read that contains one of these false softclips on the one end is very likely to have another false softclip on the other end.

[Matt Shirley]

I guess what I was asking is for you to explain the screenshot you posted. It does not seem to illustrate the question you are asking.

Answer 1

[Aerval]

Ah okay, I edited it.

We have had another look on it internally and obviously the read quality for the softclips in questions is quite bad, so that's probably the reason that BWA sets them as softclip.

[Matt Shirley]

Thanks for the clarification - now this makes sense. Glad to hear you solved your own issue.