build37 against hg19

Question

[Daisy]

Hi all

I have a raw vcf from cgatools build long time ago, from the header I can tell it is build using

##source_GENE_ANNOTATIONS=NCBI build 37.1

I used annovar to annotate it using hg19, I feel many positions has 'dot' in the annotation. I am afraid I am annotating to the wrong reference.

The raw vcf we have longtime ago (build37) , is annotating it with hg19 wrong? Again appology for naive questions, still hunting the filed :)

Thanks

[Ram]

It depends on the exact reference used. There is too little information to go off of here, but usually hg19 and GRCh37 are not compatible in annotation operations.

[Daisy]

so what can be used by annovar that matches build 37? Thanks

[finswimmer]

[...] but usually hg19 and GRCh37 are not compatible in annotation operations.

Do you mean because hg19 (UCSC) prefixed the chromomes with "chr" and GRCh37 not? Otherwise hg19 and GRCh37 should be (nearly) identical for the primary assembly.

fin swimmer

[Ram]

That could be a reason why, but the usual reason is that by the time annotation swings by, you're too far invested in one version already. I think there are reasons beyond just the name, but maybe I'm mistaken.

Answer 1

[igor]

You should spot-check the variants that get successfully annotated (in a genome browser). Do those make sense? That will tell you if the annotation is working properly.

It's possible that all the "dots" you see are non-coding. Depending on how the variants were called, it's possible that most of them are non-coding.

[Daisy]

I inspected some, they don't make sense, but don't know how to correct. I can't regenerate the raw vcf, it is old, and we don't have it raw data. But I can change annovar annotations if possible

[igor]

You can try CrossMap to convert the VCF to the desired build: http://crossmap.sourceforge.net/

[Daisy]

Oh never heard about it. Thanks thanks so much :)