csi flag in tabix_index

[FredericBGA]

I try to index a VCF that needs a .csi index.
This file can be dowloaded here:
https://usegalaxy.org/u/fredbga/h/example-vcf-gz

I didn't manage to work with the csi flag of tabix_index function.

pip freeze | grep pysam
pysam==0.16.0.1

The csi flag does not work (at least as I expected, like on command line: tabix -p vcf -C )

 >>> import pysam
>>> pysam.tabix_index('example-vcf-gz', index='example-vcf-gz.csi', preset='vcf', force=True, csi=True)                                       [E::hts_idx_check_range] Region 557988904..557988905 cannot be stored in a tbi index. Try using a csi index with min_shift = 14, n_lvls >= 6
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "pysam/libctabix.pyx", line 1035, in pysam.libctabix.tabix_index
OSError: building of index for example-vcf-gz failed
>>>

If I set the min_shift value, it works (a .csi index is created), even without the csi flag:

>>> pysam.tabix_index('example-vcf-gz', index='example-vcf-gz.csi', preset='vcf', force=True, min_shift=14)
'example-vcf-gz'
>>>

Is this the expected behavior?

[jmarshall]

This is a bug in pysam.tabix_index's csi argument handling. In this and other non-BCF cases, setting csi=True changed the filename to .csi but still wrote a TBI index (now with a misleading filename extension). Which of course does not work for wheat data.

Fixed by making the argument handling more like the tabix(1) command, for which using either -C or -m INT or both makes it write a CSI index. (TBI indices use a hardcoded min_shift of 14.)