[Enhancement] Skip troublesome reads

[YuanfengZhang]

samtools sort -@ 8 -n -m 4G 1.bam |\
samtools fixmate -cmM -@ 8  - - |\
dupsifter \
  -O dupsifter.stat  \
  ref.fa |\
samtools sort -@ 8 \
  -O bam,level=9 \
  -o 2.bam -

[dupsifter] WARNING: No valid primary alignments found for read ID: C00114:195:HV5HJDSXC:1:1101:1027:4554[dupsifter] WARNING: No valid primary alignments found for read ID: C00114:195:HV5HJDSXC:1:1101:1027:18991[dupsifter] WARNING: No valid primary alignments found for read ID: C00114:195:HV5HJDSXC:1:1101:1027:21527[dupsifter] WARNING: No valid primary alignments found for read ID: C00114:195:HV5HJDSXC:1:1101:1027:36088[dupsifter] WARNING: No valid primary alignments found for read ID: C00114:195:HV5HJDSXC:1:1101:1036:28557[dupsifter] WARNING: No valid primary alignments found for read ID: C00114:195:HV5HJDSXC:1:1101:1036:31501[dupsifter] ERROR: Can't find read 1 and/or read 2 in 2 reads with read ID: C00114:195:HV5HJDSXC:1:1101:1045:7216. Are these reads coordinate sorted?

sort -n, fixmate, collate cannot help.

[jamorrison]

For reads with the "WARNING: No valid primary alignments" message, these reads will not be marked as duplicates and the code will move on to the next read. I can look at modifying the --verbose option to take a level value that will allow WARNINGs to not be printed if the user requests that though.

To the error, this usually means the reads are either coordinate-sorted or there is a mix of single-end and paired-end reads in the BAM (or that a paired-end read is missing its mate entirely). Since sort -n and related options aren't working, I'm guessing there's a paired-end read missing its mate. I'm not sure why that might be, but that would be a first place to look. Let me know if that seems to be the issue. If so, I can give you a couple options for resolving that issue. Otherwise, if you could send a small example that recreates the issue and I'll take a look into what may going on.