dalexander/htslib_tell_issue

Can we get the offset of the first alignment within a range?

The answer seems to be “no”. Building the iterator (sam_itr_queryi) does not seek, and the first time you use advance the iterator (bam_itr_next), you are already past the first read.

Here are the offsets of the first reads in the file, along with their [tStart, tEnd) spans:

./htslib_read_tell bam_mapping.bam | head -10

#+RESULTS[42d729a36eb2cca2afa3f0e29c4368acb3d4b55b]:

0x2390000               0       471
0x2391d38               302     1019
0x2393275               675     1026
0x2393c78               2168    2326
0x23941d3               2170    2397
0x23949a8               3572    5015   # This is the first read in our query window
0x2397209               4506    6125
0x239a0e2               4507    5860
0x239c73e               4592    5203
0x239d92b               4658    5011

Now here’s what happens when we try to query the range [5000, 6000)

./htslib_itr_tell bam_mapping.bam

#+RESULTS[dbc566df424a2a721ebc5917e95e8486b37271b3]:

0x2390000               3572     5015   # Note incorrect offset here, from pre-seek
0x2397209               4506     6125
0x239a0e2               4507     5860
0x239c73e               4592     5203
0x239d92b               4658     5011
0x239e3ba               4842     5215
0x239ef0c               4903     5380
0x537e0000              5135     5388
0x537e07da              5426     7039
0x537e353a              5942     6134