tabix with specified coordinates

Question

[qmemcm]

Hi!

I am using tabix to output from a vcf.gz file, some regions of interest. What is interesting is that when I specify, say,

10:7073963-7077040, I get results from regions outside of the range also. For example I get something at position 7073962. Why does tabix do this? Is this a bug? Or is this a built-in feature?

Please let me know.

Thank you.

[Devon Ryan]

Is it a SNP that you're getting at 7073962 or a longer InDel? The latter would overlap your specified range.

[Pierre Lindenbaum]

I don't think tabix looks at the bases when indexing. It only uses CHROM & POS.

[Devon Ryan]

True, you're probably correct then that this is a 0 vs. 1-based issue.

[Pierre Lindenbaum]

It's not clear to me if tabix always uses a 0-based coordinate system.

Answer 1

[qmemcm]

I think Devon's original answer is correct. I just gave that one example. But I have another: I specify the region as 721960-724370, and I get a long indel starting at 721929, which certainly overlaps my region.

Any more thoughts?

Thanks!

[Devon Ryan]

I should really refresh my memory of the tabix code, that'd answer this straight away. I know that there's code to deal with starts/ends, since a tabix index is basically an R-tree for block gzipped files. The question simply becomes whether the parser handles VCF files in an intelligent way or if it just blindly assigns start and and coordinates based on the same VCF column.

Edit: Note that if what I wrote above makes no sense to you that that's not something to worry about :)