Pysam ignoring some reads?

[rLannes]

Hi,

First a bit of context:
I do want to modify TEtools to be able to use pysam so it can accept bam.

issue:
But my version returns less read than the TE tools version.

Can someone explain to me why my code returns fewer reads? From my understanding, they are equivalent, except for the use of pysam. Note my version returns the exact number of mapped reads (samtools flagstat), TEtools returns more. (does pysam filter secondary per default?)

TEtools version:

with open(self.sam_file) as sam_file_handle:
   for line in sam_file_handle:
      if line[0] != '@':
         line = line.split()
         if len(line) > 4 and line[2][0] != '*' and int(line[4]) <= self.max_mapq:
            self.rosette.count(str(line[2]), 1)

My version:

with ps.AlignmentFile(self.sam_file, open_mode)  as sam_file_handle:
   for aln in sam_file_handle.fetch(until_eof=True):
      if aln.is_mapped and aln.mapping_quality <= self.max_mapq:
         self.rosette.count(aln.reference_name, 1)

Any help warmly appreciated.

Romain,