flagstats not equivalent to samtools

[drmjc]

Hi,
thanks for putting together an excellent toolkit, the multithreading is a real win!

I ran samtools flagstat, and then sambamba flagstat, and noticed that they handle secondary alignments differently. It seems sambamba doesn't account for secondary alignments, and then overestimates 'paired in sequencing' by the number of 'secondary's; likewise the sum of differences in read1, read2 = the number of 'secondary's.

Would you consider either documenting this on the readthedocs page, or be able to look in to making these behave the same?

samtools output

25678848 + 0 in total (QC-passed reads + QC-failed reads)
9066 + 0 secondary
0 + 0 supplimentary
0 + 0 duplicates
25606121 + 0 mapped (99.72%:-nan%)
25669782 + 0 paired in sequencing
12834891 + 0 read1
12834891 + 0 read2
25455674 + 0 properly paired (99.17%:-nan%)
25525802 + 0 with itself and mate mapped
71253 + 0 singletons (0.28%:-nan%)
32460 + 0 with mate mapped to a different chr
22265 + 0 with mate mapped to a different chr (mapQ>=5)

sambamba output

25678848 + 0 in total (QC-passed reads + QC-failed reads)
0 + 0 duplicates
25606121 + 0 mapped (99.72%:-nan%)
25678848 + 0 paired in sequencing
12839056 + 0 read1
12839792 + 0 read2
25460539 + 0 properly paired (99.15%:-nan%)
25534779 + 0 with itself and mate mapped
71342 + 0 singletons (0.28%:-nan%)
37768 + 0 with mate mapped to a different chr
24202 + 0 with mate mapped to a different chr (mapQ>=5)

[lomereiter]

Hi,
Thanks for pointing this out. I found the two samtools commits where the behaviour was changed, will make it identical in sambamba shortly.