hisat2 output more reads than are in the file

[dpryan]

52.68% = (2*(4647893+779552+138165) + 510907 + 357302)/(2*11389273)

Samtools is correct, since if sum the numerator you'll get just under 12 million total alignments, which is exactly what's in the BAM file (14328481-2329052).

[frymor]

Hi all,
I am in the middle of testing the hisat2 mapper and encountered a discrepancy between the output hisat2 gives me at the end of the mapping step and the number of reads samtools flagstat counts.

this is the output I get, when hisat2 is finished:

cat ../hisat2Mapping/WCE7.stat 
[B]11389273[/B] reads; of these:
  11389273 (100.00%) were paired; of these:
    5961828 (52.35%) aligned concordantly 0 times
    4647893 (40.81%) aligned concordantly exactly 1 time
    779552 (6.84%) aligned concordantly >1 times
    ----
    5961828 pairs aligned concordantly 0 times; of these:
      138165 (2.32%) aligned discordantly 1 time
    ----
    5823663 pairs aligned 0 times concordantly or discordantly; of these:
      11647326 mates make up the pairs; of these:
        10779117 (92.55%) aligned 0 times
        510907 (4.39%) aligned exactly 1 time
        357302 (3.07%) aligned >1 times
52.68% overall alignment rate

@Question - how are the 52.68% are calculated? what reads are being considered here as mapped?

and this is the number of reads I get, when I run samtools flagstat on the sorted/indexed bam file:

samtools flagstat ../hisat2Mapping/WCE7.sorted.bam
[B]14328481[/B] + 0 in total (QC-passed reads + QC-failed reads)
2329052 + 0 secondary
0 + 0 supplementary
0 + 0 duplicates
14328481 + 0 mapped (100.00%:-nan%)
11999429 + 0 paired in sequencing
6098403 + 0 read1
5901026 + 0 read2
10854890 + 0 properly paired (90.46%:-nan%)
11407498 + 0 with itself and mate mapped
591931 + 0 singletons (4.93%:-nan%)
23248 + 0 with mate mapped to a different chr
18678 + 0 with mate mapped to a different chr (mapQ>=5)

As you can see, samtools find more reads than there suppose to be originally in the file.

Is there a simple explanation for that?

thanks,
Assa