Not 100% mapping in output bam?

[ronaldrcutler]

Running samtools flagstat on a bam file resulting from alignment with Hisat2 gives:

76075665 + 0 in total (QC-passed reads + QC-failed reads)
1565341 + 0 secondary
0 + 0 supplementary
0 + 0 duplicates
71435955 + 0 mapped (93.90% : N/A)
74510324 + 0 paired in sequencing
37255162 + 0 read1
37255162 + 0 read2
64430312 + 0 properly paired (86.47% : N/A)
67187398 + 0 with itself and mate mapped
2683216 + 0 singletons (3.60% : N/A)
2452092 + 0 with mate mapped to a different chr
2095660 + 0 with mate mapped to a different chr (mapQ>=5)

Shouldn't there be 100% mapping? I think because of this I also get the warning described below.

I also notice this warning at the end of an HTSeq-count log file when counting features in the same alignment file above:

Warning: 284233 reads with missing mate encountered.

Should I be worried about this warning? Note that the same raw reads aligned with Tophat2 had 100% mapping in the respective bam file.

[dpryan]

The flagstat output will only show 100% mapped if the aligner doesn't output unaligned reads. Tophat doesn't output unaligned reads, hisat2 does, so this value is meaningless on tophat output.

The htseq-count warning is probably due to an incorrect flag being set in tophat's output (tophat always had the problem, so I wouldn't be surprised if hisat2 does as well). As an aside, featureCounts is vastly faster than htseq-count.

[ronaldrcutler]

Thanks Devon, helps a lot. I'll make sure to give featureCounts a go.