Tabix doesn't handle queries using sequence names containing ':' or '-'

[cmnbroad]

TabixReader parseReg method assumes these characters don't appear in sequence names. This came up in the review for #911.

[lindenb]

@cmnbroad

FYI: for VCF, the spec says that a colon shouldn't appear in the CHROM.

in https://samtools.github.io/hts-specs/VCFv4.2.pdf

The colon symbol (:) must be absent from all chromosome name

handling a colon in a vcf.gz would be wrong isn't it ?

[droazen]

@lindenb In the hg38 reference we find contigs such as this one:

@SQ	SN:HLA-A*01:01:01:01	LN:3503	M5:01cd0df602495b044b2c214d69a60aa2	AS:38	UR:/seq/references/Homo_sapiens_assembly38/v0/Homo_sapiens_assembly38.fasta	SP:Homo sapiens

[yfarjoun]

This is due to an inconsistency between SAM and VCF specs regarding what constitutes a legal contig name.... It would be good to solve this at the spec-level...but in the meantime...we needs a solution. :-/

…

On Mon, Oct 23, 2017 at 11:14 PM, Pierre Lindenbaum < ***@***.***> wrote: @cmnbroad <https://github.com/cmnbroad> FYI: for VCF, the spec says that a colon shouldn't appear in the CHROM. in https://samtools.github.io/hts-specs/VCFv4.2.pdf The colon symbol (:) must be absent from all chromosome name handling a colon in a vcf.gz would be wrong isn't it ? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1017 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACnk0hXRvZ16cjz1OsizhKnkC2vSjKTSks5svPOOgaJpZM4QDa60> .

[jkmatila]

Would one possible workaround be to add support for escaping the problematic characters in the chromosome name in tabix when parsing the region specifier?

For example, quoting the problematic characters by prepending a backslash, like HLA\-A*01\:01\:01\:01.

Or alternatively, using the percent encoding, like HLA-A%2A01%3A01%3A01%3A01.

[jmarshall]

Since 2017, both SAM and VCF have gained clarity around what constitutes a legal contig name (and VCF's prohibition of colons has been removed). The SAM spec also has some suggestions about parsing region specifiers in the face of extra colons: see Appendix A of SAMv1.pdf.