Calculation of split-read end position in certain cases is off

[dminkley]

Hello,

After making the changes that address issue #18, I noticed that the first-occurring portion of some split reads indicated different SV endpoints than the other SVs (they generally ended in staggered positions within the DEL SV itself). Looking in to this, this portion of the add_split() function seems to be the problem:

            qs_pos, qe_pos = \
                    calc_query_pos_from_cigar(cigar, strand)
            splits[sr.HP][read.query_name].append(SplitRead(pos,
                                                  pos+qe_pos,
                                                  strand,
                                                  qs_pos))

pos+qe_pos is currently used to identify the endpoint of a split read fragment - this isn't always correct. For the first-occurring split read portion of reads on the negative strand, this will be wrong. Changing it to pos+(qe_end-qe_pos) seems to do the trick!

The example I used in issue #18 also demonstrates this issue.

-David