Intronic Snp Functional Annotation

Question

[Varun]

Hi!

I've performed a candidate gene association study and my top SNPs are all intronic SNPs. I've run them through a couple of basic annotation software online, checking for the usual stuff as well as some of the more recent things like miRNA binding sites etc. Two of my top hits are very close to each other ( <1000 bases) but aren't in high LD. The only thing I can think of, that could possibly result in the intronic SNPs causing a change in the phenotype, is that they are in LD with some other variant, but a trawl through the normal databases hasn't proven fruitful. Does anyone know how I can check for variants that are in LD with my SNPs? Any ideas on how else I can functionally annotate my SNPs ?

Secondly, all my top SNPs are in repeat-rich regions. Apart from the fact that repeat rich regions are recombination hotspots, any other idea of how this might be significant? Any software that I can use to examine this further?

Thanks!

Answer 1

[Noel]

Hi, by "basic annotation software online", I suppose you have looked for insulators, CTCF binding sites, transcription factors, etc.? Additionally, is it a CpG? I normally use UCSC's browser to see in which LD block my SNP's are located (last time I checked the LD track was only available for hg18).

Answer 2

[Larry_Parnell]

With which organism are you working? Repeat elements in different organisms have different functions. Some harbor (very strong) promoter elements, some insulators (mentioned by Noel). It is also possible that your SNPs tag a rare variant that you have not genotyped.

Do any of the SNPs or those in LD show eQTL associations?

It seems that you are after computationally predicted function in lieu of a functional assay, but the latter will give your results more impact. Nonetheless, if the SNP is in a repeat, one can assess how much variation is tolerated at other instances of that repeat element elsewhere in the organism's genome.

Lastly, intronic SNPs can map to splice enhancers. Have you tested for this?

Knowing precisely what you've checked and what those checks produced would be helpful in allowing us to help you more precisely.

Edit added 26 Dec 2012: Use the eQTL browser at Univ of Chicago to check for eQTL associations.

Answer 3

[Varun]

Thanks, Noel. I tried UCSC's genome browser earlier, but haven't found anything significant.

Larry, I'm working with humans. We aren't in a position to do functional tests, due to the nature of the phenotype and genes which have emerged significant. The criteria I've tested my SNPs for are: Effect on coding sequences, miRNA, snoRNAs and scaRNAs based regulation, CpG islands, transcription factor binding sites, splice enhancer, splice acceptor and splice sites, non-coding human enhancers, DNAse I hypersensitivity, recombination hotspots, Tajima's D value. I've done this with the significant SNPs and SNPs in LD (with an R-square of above 0.8).

How do I check for eQTL associations? Further, do you have any suggestions as to how I can check for tolerance in variance? But wouldn't this also be dependent on the position of the repeat in the genome?

Thanks!

[Larry_Parnell]

See my edit added to my response regarding eQTL data.

[Varun]

Thanks, Larry - very helpful!

Answer 4

[Erik Garrison]

I believe the variant effect predictor provides some annotation of intronic variants.

If you have sequencing data, you could also try to detect indels and structural variations. Although they are more numerous and much easier to observe, SNPs don't have the same degree of functional effect that larger sequence variants do.

Answer 5

[Varun]

Thanks, Eric. I'll check variant effect predictor. Unfortunately, my study is only a genotyping study, so don't have sequencing data. Yes, I agree - SNPs don't have a great degree of functional effect.