If i am interested in a transcription factor, why would i focus on coding and non coding SNPs?

Question

[mesencephalon]

I have recently been reading about transcription factors and SNPs and was the above question came to my mind.

What kind of impact would a coding or a non-coding SNPs have on transcription factors? Is it dependent on the location of the SNP?

What kind of impact would it have if the SNP is located on a promoter, enhancer, intron and intron/exon boundary?

Answer 1

[Emily]

In short, yes, non coding SNPs do have an impact. I recommend reading up on transcription factor binding motifs and eQTLs. Here's a nice paper: http://www.nature.com/nrg/journal/v16/n4/full/nrg3891.html

[ivivek_ngs]

Very rightly said. Just to add up, unless one performs the eQTL studies you will not be able to correlate the impact of variation. The idea is simple if you have a protein-coding variation then if you see that the transcript it targets can directly have an impact on the gene regulation depending on the type of variation.

When you annotate your variants and you find them in non-coding regions, remember they might span the boundaries in intron-exon or intergenic or UTRs. Then if you perform eQTL studies and try to find the cis-regulatory domains you will be able to understand if they will still impact regulation of the gene or not. Now having access to eQTL databases also let you know if there are any TFs that bind to that location or not. Also there are other ways to assess that and you will be able to obtain that , then one can also use their own experimental designs in vitro to perform target TF enrichment for those specific TFs that you find through you eQTL studies mapping non-coding variants to regulatory regions.

Some papers:

Link1

Link2

Answer 2

[Denise CS]

I remember from a talk I gave with Daniel Zerbino that only 5.3% of variants from the GWAS catalog were located in coding sequences/splice sites. Intronic and intergenic variants accounted for 46% and 37%, respectively. In addition, non-coding regions can be extremely conserved across vertebrates and some of them can be bound by TF such as CEBPA.

So no wonder there is such hype on non-coding regions. They must (ought to) have an impact somehow. The 'how' is the tricky bit though.

There are many examples in the papers suggested by @Emily_Ensembl and @vchris_ngs, and the FTO is a classic one for me. FTO not only have an intronic variant associated with obesity, but the effect of this variant is not on the FTO, rather another gene.

For more on eQTL, functional genomics, LD, and assigning a variant (coding or otherwise) to a gene associated with a disease, check this paper out.

[ivivek_ngs]

I thought of putting LD association studies as well based on GWAS but then that would have come up with specific queries from OP. Nicely written.