Where Can I Find Information Related To A Snp That I Think Is Responsible For Disease?

Question

[Mel]

I have a SNP I think is responsible for disease. What database can I use to see if there is an association of this SNP with disease? My disease of interest is diabetes. Alternatively I would look at other parameters like obesity.

I would like to test my hypothesis with a quick and dirty snapshot before entering into a more detailed professional analysis.

Thanks for any help.

M

[Neilfws]

Just a reminder that diabetes is not one disease; there is type-1 and type-2.

[Pierre Lindenbaum]

See the also the question "Disease associated SNPs".

[Larry_Parnell]

Some have even proposed a Type 3 diabetes.

Answer 1

[Jorge Amigo]

this is indeed a very common need. in fact important that large projects such as the Human Variome Project are trying to solve. not only having a database containing mutations detected is interesting, but also the curation of their phenotype is critical in order to fully trust in the databases conclusions.

right now, there are 2 accepted approximations that you may go for: either check for your SNP on a large general purpose database (phenotype information would be absent or limited), or either look for it on a locus specific database (phenotype description and pathogenicity should be better described). both approximations are valid, and if you don't need to be 100% sure of their predictions but to get "a quick and dirty snapshot" you will find useful any of them.

although there are plenty of databases you may want to query, and sure upcoming answers will add more options here, the ones we normally tend to use are dbSNP, HGMD and HGVbaseG2P (now GWAS Central), all 3 from the general purpose group. my group is not dedicated to diabetes, so all I can give you is a LSDB I know that describes diabetes mutations quite well: the LOVD on monogenic diabetes. I don't know if that's the one you're interested in, but you will find more than 8000 variants (almost 3000 of them unique) reported there.

[Jorge Amigo]

the main problem I see is that we bioinformaticians try to use the best available databases (and the larger we see they are, the more we like them), but what we consider best is not what clinicians may need. in my group we have several disease oriented research lines (cancer, neuro, psych, ...), and they all prefer to deal with a few hundreds/thousands mutations very well described that they could blindly trust, rather than using a very dense database where the phenotypic data won't necessarily be completely trustworthy. I guess somewhere in between is where we should all meet.

[Chris Evelo]

Jorge is absolutely right abou the phenotypic information being important. And it is in fact hard to get. After working in the nutrigenomics field for a few years I came to realize that it is harder to determine what they eat than what the content of heir genome is. This is why initiatives like HVP are so important.

[Chris Evelo]

Jorge is absolutely right about the phenotypic information being important. And it is in fact hard to get. After working in the nutrigenomics field for a few years I came to realize that it is harder to determine what they eat than what the content of heir genome is. This is why initiatives like HVP are so important.

[Chris Evelo]

Jorge is absolutely right about the phenotypic information being important. And it is in fact hard to get. After working in the nutrigenomics field for a few years I came to realize that it is harder to determine what people eat or how healthy they are than to find out what the content of their genome is. This is why initiatives like HVP are so important.

Answer 2

[Chris Evelo]

The database that per disease (actually diabetes is more than one) and per gene describes what that the known relations is [?]OMIM[?]. Now with the rapid collection of information about individual SNPs associated with disease, OMIM gets out of sync, and the other databases become more important. But OMIM will still show you what we really know. OMIM is intended more as a human readable than as a computer readable database. But if you really know the disease you are interested in it might not be a bad idea to start there, since you will have to read up anyway.

SNPedia is a community curated resource. Here you can check an individual SNP to find out whether there are known strong relationships with disease. You can also [?]search it for diabetes[?].

Don't forget to also do a Pubmed search. New information on this subject is collected at such a speed that not everything might be in the databases yet.

If you do a more structured (some people would say more real bioinformatics) search, using for instances the databases that Jorge suggested, and present your final results to biologists or doctors, please don't forget to also give them links to the resources I mentioned here, since even if you want to skip the reading they will want to do it.

[Jorge Amigo]

whoops... forgot OMIM on my answer ;) probably one of the first places I would start from.

[Neilfws]

+1 for searching SNPedia, where there seems to be plenty of easily-accessible information.

[cdsouthan]

One needs to be circumspect about OMIM (or MIM as many now call it) in particular the difference between a high penetrance Mendelian disease with data-supported molecular mechanistic causality and disease association results where we mostly dont have even a mechanistic clue. Also, for reasons unclear to me at any rate, MIM indexes 1000s of human proteins with no disease-linkage evidence. Adressing this question would be much more useful if the poster could tell us what the SNP was! They can put it up somewhere else first (their blog, GitHub or figshare) to stamp their provenance and attribution. And they can write the paper long before anyone reading this could (who would not anyway) and we could help out for sources.

N.b. Just noticed this was three year old post! - caught by the moderation updates again, but no harm in leaving my comment I guess

Answer 3

[Maximilian Haeussler]

Don't restrict yourself to Pubmed, rather try a fulltext search like Google Scholar. The UCSC genome browser also has a track with rsXXXXX matches in fulltext.

Other general Databases: GAD, ClinVar, UniProt (for coding mutations)