pileup reloads BAI index

[suhrig]

Hi,

I use the pileup() function in a loop to generate pileups over a number of genomic ranges. pysam appears to reload the BAM index for each invocation of the pileup() function even though I'm using the same AlignmentFile object. Is this intended behavior? I creates considerable overhead.

Steps to reproduce:

1. Create a minimal example python script:

#!/usr/bin/env python3
import pysam
bam = pysam.AlignmentFile("input.bam", "rb")
for i in range(100):
    for pileupcolumn in bam.pileup("1", 1000, 1010):
        pass
bam.close()

1. Run the script with strace to log file access:

strace -o strace.log ./example.py

3. Count how often it opened the BAI file:

grep -c 'openat.*[.]bai' strace.log

Result: 101 times