Insertions counted twice (in STAR) + Default MM set to <4

[eyalmpeer]

Hello. Great program and article.
Two possible issues:

1. STAR aligner's NM tag includes insertions (in contrast to the "nM" tag which only includes mismatches in each pair).
   When calculating the MM score XenofilteR, at least when used on STAR aligned files, possibly counts insertions twice:
   Once from the CIGAR field, and once when adding the NM tag.
   So, a read which has two insertions and no mismatches will be filtered out because the MM score will be 4.
2. Another issue is that the default MM_threshold is documented as 5 but in practice in script it is 4.
   Because the script only includes reads with MM score < MM_threshold (and not <=), all MM_score 4 or above will be filtered out, when you may have intended that only 5 and above will be filtered out.

Thanks!
Eyal

[RoelKluin]

Thanks. I'll add a NM_id argument, defaulting to "NM", but which can be adapted for STAR. Also the mismatch value was indeed not reflected correctly in the comments and README, I've adapted those to better reflect the mismatch threshold in use, hope this covered all occurences - the actual value in use by XenofilteR wasn't changed. BTW, If one expects clipped reads it may be beneficial to disable the threshold entirely by setting it to a value higher than the read length.