Annotate fail to add INFO when match structural variation by POS,~ID,REF,ALT

[Han-Cao]

Hi,

I am using bcftools 1.15.1. I want to add INFO for a vcf file of structural variations.
The vcf is like:

chr1	10627	ID1	N	<INS>	.	PASS	some_INFO
chr1	90238	ID2	N	<INS>	.	PASS	some_INFO
chr1	90337	ID3	N	<INS>	.	PASS	some_INFO
chr1	90388	ID4	N	<INS>	.	PASS	some_INFO
chr1	90412	ID5	N	<INS>	.	PASS	some_INFO
chr1	90416	ID6	N	<INS>	.	PASS	some_INFO

The annotation file is like:

chr1	10627	ID1	N	<INS>	-0.999642
chr1	90238	ID2	N	<INS>	1.25927
chr1	90337	ID3	N	<INS>	1.42278
chr1	90388	ID4	N	<INS>	1.14841
chr1	90412	ID5	N	<INS>	1.19886
chr1	90416	ID6	N	<INS>	-0.556859

When I first try to annotate by POS,~ID,REF,ALT suggested by the manual (bcftools annotate -a annots.tab.gz -c CHROM,POS,~ID,REF,ALT,INFO/VAL input.vcf), no value can be added to the INFO.

However, after I remove the REF and ALT columns in the annotation file and annotate by -c CHROM,POS,~ID,INFO/VAL, I can get the expected output.

By the way, I first try to annotate from vcf, where I found todo: -c ~ID with -a VCF?, does it mean this feature is on the todo list and currently we can only annotate by tsv file?

Thanks,
Han

[pd3]

Hi, yes, that's correct. This has not been done for the case when annotating from a VCF yet

[tobiasrausch]

Matching with symbolic ALTs or based on IDs would indeed be very useful for structural variants: bcftools annotate --pair-logic id ...

Thanks!

[maggs-x]

Hi. I have a vcf that has SNPs and structural variants. The vcf format is very simple. There is no type in the INFO column and we are inferring structural insertions and deletions based on the length of the ALT allele compared to the REF. I noticed I get differing results from bcftools annotate depending on if I annotate a vcf that has multiallelic sites joined vs a file that does not have multiallelic sites joined. For example, there are fewer genes annotated in the vcf that has multiallelic sites joined.

I've been unable to determine if the problem is related to how 'bcftools annotate' handles multiallelic sites, or if it is how it handles structural insertions and deletions (50bp-100kbp). Please let me know. Thank you.

Maggs

[pd3]

@tobiasrausch The work is done, see the pull requests samtools/htslib#1844 and #2298

[tobiasrausch]

Great, thanks Petr!