How does bedtools - bamToBed handle S operation in cigar string?

[RedLightPanic]

Hi,

I've been recently using bamToBed utility from the bedtools suite in order to transform the alignments in a BAM file into features of a BED file.

I have one question regarding the way bamToBed handles the "S" operation in the cigar string.

For example, I have this alignment (in BAM/SAM format, obtained using STAR mapper. I'm omitting certain not relevant parts of the alignment):

HANNIBAL_0054_FC62M9BAAXX:7:53:2326:12457#0/1   0   chr1   28166   251   4S25M7S   *   0   0   ...

that bamToBed transforms to the following BED feature:

chr1    28165   28190   HANNIBAL_0054_FC62M9BAAXX:7:53:2326:12457#0/1   251     +

I was just wondering why the BED feature is 25 nts. long (25M), and starts at the same position of the BAM alignment. Considering that the first operation in the cigar string is 4S, shouldn't the BED feature start 4 nucleotides after the position of the BAM alignment?

Thanks a lot in advance.

[dpryan]

The start position in the SAM file refers to the 5th base of the read in your example. Soft-clipped bases don't offset the start of the mapped position in SAM.

[RedLightPanic]

Than you so much Devon.

Didn't find this in the sam specs.

[dpryan]

Yeah, the SAM spec is unfortunately not always very explicit. This is only covered in an example on the first page rather than the main part of the spec.