Assign Intergenic Snps To Gene.

Question

[Peixe]

Hi,

This is a trivial question and I am surprised I could not find it anywhere else...

How can you map an intergenic SNP to a unique gene? I mean, is there any "canonical good way" to do it? Several ways to do it come to my mind: the closest in distance, the one with the higher R² value between the SNP and the first gene polymorphic position, etc... But I would like to know if any previous reference of someone doing it somehow do actually exist. GWAS usually report both genes (upstream & downstream), but I'd rather like to have a single hit.

Does anyone know a good way to do it?

[Giovanni M Dall'Olio]

I don't have time to write a full answer now, but have a look at this paper: Habegger et al, 2012. VAT: a computational framework to functionally annotate variants in personal genomes within a cloud-computing environment.

[Giovanni M Dall'Olio]

In this recent paper Raj et al, 2013, Common Risk Alleles for Inflammatory Diseases Are Targets of Recent Positive Selection they assigned SNPs to the closest gene inside the LD block.

[Peixe]

I know... & you know we are doing a journal club on it soon... ;) hehehe...

Answer 1

[Larry_Parnell]

Intergenic SNPs are likely to either affect transcription (mapping to a promoter or enhancer of a nearby gene) or map within a currently undescribed/unknown gene (eg, a novel lncRNA). One approach is to see if the SNP, or an LD partner SNP, has eQTL properties with a nearby gene. These data (from GenVar and UChicago eQTL tools) are mostly for protein-coding genes. If you're far from such genes, you may be in an enhancer and that's tougher to discern function and assign a relationship to a gene. There are enhancer tools and databases but showing the link to a gene controlled by that enhancer will be more difficult to obtain. Lastly, some non-coding RNAs might show sequence conservation across species but not always, and not at the precise position upstream/downstream of some anchor, like a protein-coding gene - which all means it could be difficult to say that your SNP maps to a gene encoding a novel non-coding RNA.

More to the point of your question: The Framingham Heart Study, which has initiated many GWAS, has used a distance to 60 kbp to assign a SNP to a gene. A distance greater than this is not assigned to a gene or done so but treated as "distant." You can always find the nearest gene, or one with an NM_# RefSeq mRNA (as opposed to a gene model mRNA), and the distance to that gene - carry forward both values.

[Peixe]

That's a nice explanation, @Larry_Parnell ! My SNPs are mostly derived from GWAS, so think I'd go for the "nearest gene" criterium. However, I think is worth to give a look to the Genevar. Thanks!

[Larry_Parnell]

Thanks you. You should use as many eQTL resources as you can find. Estimates range from 30% upwards that GWAS hits affect transcription and so eQTL analysis will be very helpful for you.