Handle no-ops more intelligently when creating MD tags

[kristalcurtis]

Currently, Adam creates very verbose MD tags when it encounters no-ops, e.g., 2G0A0T0G0A2G2A0T0G0T1G0T0C3T0G0A0G0T1G0T0C0A1G0T0C0T0G0A0T0G1A3G0A0C0A0T0C0A0T1G2C0A0G0A0T0G0C0T0G0A0G4C0A0G1C0A2C0A0T0A0T1G0T0G9.

[fnothaft]

D'oh! That doesn't look correct... Which piece of code is emitting this?

[fnothaft]

Er, actually, if those all are mismatches, that could be correct. Are you able to share the read/alignment publicly?

[kristalcurtis]

Sure, I can share; it's just from the SMaSH Venter reads.

Here's the original read:
@chr22_42898209_42898675_?:?:?_?:?:?_MATERNAL_42933514_42933981_649d432/2
GGGGGGGGGGGCACCATATGGGTGGCTGGGGGCTCAGCATCTGGGCCATGATGTCCCCTTCATCAGACCTGACCACTCAAAAGACCACATTTCCCTCATCC
+
CGFGGG?FGGCGFGGAGG4GFGGG6FFAGGGGDGEFFGFGGGDB@GFGGGGG?GGGGGFFEEFFGG@>EGGBGGGGG@G16?DGB4GDGFBEGFEGFFGF>

It aligns to chr22, 42898574 (if 0-indexed), in the reverse direction. Here's the cigar string & MD tag I get from Adam:
cigar: 10M1I1M2I1M1I3M1D3M1D5M1D4M1I4M1I4M2I1M2I16M4D4M1D5M1D2M1I1M1I9M1I3M1D1M2D1M1D9M
mdTag: 3A0T1A3A0A0A0T0G0^T2T0^C1T0T0T0G0^A1T2T0C0A0G0G0T0C0T1A0T0G0A1G1G0G0A1A1C0A0T0^GGCC0C0A0G0A0^T0G1T0G0A0^G0C1C0C2G0C0C0A0C1C0A0T0^A1^GG1^G0C3C1C0C1

It looks like there's an indel at the beginning of the read:
scala> genome.substring(42898574, 42898574 + 101)
res6: String = GGGATGAGGGAAATGTGGTCTTTTGAGTGGTCAGGTCTGATGAAGGGGACATCATGGCCCAGATGCTGAGCCCCCAGCCACCCATATGGTGCCCCCCCCCC

Maybe that's what is causing the hiccup? I realize this MD tag is different from the one I posted... maybe the code changed in between me getting the above result (about a week ago) and now?

[fnothaft]

Hmmm, that's a messy alignment. I'll look into this...

[fnothaft]

Closing as won't fix.