Use cs tag about sniffles HOT 7 CLOSED

Minimap2 will be updated to include MD tag. Thus this is getting irrelevant.

from sniffles.

Thanks much for the explanation, @philres. I will consider to implement this tag but not very soon as @fritzsedlazeck thinks the MD tag alone should work in most cases.

from sniffles.

I didn't know Sniffles requires MD to work; otherwise I would have --MD earlier. Please let me know if Sniffles requires more SAM features. I'd love to make minimap2 work with it. Thank you!

from sniffles.

Thanks. Its scans the MD for "small" INDELs and thus dont operate on the reference sequence.

If you include the MD string that is 98% of all the magic. Philipp and I also included a SV tag in NGMLR that indicates for me if a read was truncated because of N's or because of SVs. Happy to explain.

Thanks.
Fritz

from sniffles.

Could you give a bit more detail about this SV tag? Thanks.

from sniffles.

Sure. Happy to discuss also offline. SV is a binary tag in NGMLR to indicate the reason for clipping.
@philres knows more about the details:
philres/ngmlr#15

This is used to infer larger Ins than read length.

Thanks
Fritz

from sniffles.

Hi Heng,

The SV:i tag in the NGMLR output contains two bit flags:

0x1 indicates whether the reference sequence consists of mostly Ns up and/or downstream of the read alignment. Sniffles uses soft-clip information to improve detection of (large) insertions. As a consequence longer regions of Ns in the reference would cause false positives signals as the reads up/down stream of the Ns will all be soft-clipped at the same position of the reference. 0x1 helped to avoid that.
0x2 is set if > 95 % of the read base pairs were aligned in the particular alignment. Sniffles could compute that from the CIGAR string but since we had the SV tag already at that point we just added it. Don't know if Fritz is still using it.

Hope that helps,
Philipp

from sniffles.