HLA analysis using SNP2HLA

Question

[bioinf]

Hello

I have run snp2hla in my dataset and got the imputed results. But I am unable to interpret it. The softwares website also does not have sufficient information. Could anyone help me on this. Thanks in advance.

[Titus]

You should show your output to get help :)

Answer 1

[jyoti.khadake]

The output has P or A against the haplotype which indicates either presence or absence of haplotype

Answer 2

[Charles Warden]

I would have to couple check the output files on my home computer, but I think there should be a way to define your top 2 haplotypes.

For example, I have these comparisons with different HLA programs for my own genome data (underneath the huge "23andMe versus Genes for Good" Venn Diagram):

https://github.com/cwarden45/DTC_Scripts/tree/master/Genes_for_Good

For example, concordance was generally better for my HLA-A, HLA-B, HLA-C genes.

[Shicheng Guo]

Hi Charles, Soumya mentioned in his Nature Genetics paper 2012. 2767 SNPs were selected to tag the entire MHC. Where I can find the list for these 2767 SNPs. Thanks.

[Charles Warden]

That is a good question - I am not entirely sure (which sites are most informative for the SNP chip imputation, or which ones are used in that paper).

With the NGS data, you'd probably want a specific HLA reference with the HLA haplotypes (rather than a more typical genome alignment).

If you look up HLA-A on the UCSC Genome Browser, you get a lot of alternative chromosome coordinates. I don't remember seeing those as often on SNP chip data, but perhaps you could look in the range of variants for the overall HLA region on chr6?

I'll double-check where I see a dip in my Veritas WGS data when I get home (since I believe they provided a filtered bam per chromosome, instead of raw .fastq files). So, perhaps this can be revised, but maybe sites within chr6:32,390,512-33,117,623 (on hg19) could be of interest?

Update: My Veritas WGS data has a dip from chr6:28,478,344-33,451,189 (for reads that weren't aligned to the canonical chromosomes, and not having unaligned reads to test for re-alignment). That is a larger region than I originally guessed. I suspect there are also regions mapped to other areas (possibly with homology to the alternate HLA chromosomes), but this was the most clear to me.