[jebus989]

Of course, they're just calling SNPs and using published GWAS studies to estimate disease risk. You can get your "raw genome" text file from 23andMe and convert to e.g. VCF format for use in a bioinformatics tool like the variant effect predictor.

Besides, really the most interesting health-related alleles are the simplest: ApoE, BRCA1 — no complicated algo is needed to interpret those associations.

[bentcorner]

Thanks for the comment - my wife and I submitted 23andme kits for fun and learned some of our history, but I was disappointed we weren't able to get any health information from it. I didn't know you could get your genome info from 23andme and get it interpreted by someone else!

[jessriedel]

> You can get your "raw genome" text file from 23andMe

Well, not your full genome. You can get the raw SNPs that 23andMe test for, and if you're in their pilot program that sequences the exome (which is a superset of the SNPs but a subset of the full genome) then you can presumably get the raw data on that.

[jebus989]

That's their terminology I believe (hence quotes). However, for fun you can have a go at imputing a full genome from these SNPs (e.g. http://genomesunzipped.org/2013/03/learning-more-from-your-2...). SNPs aren't necessarily exonic either so exome-seq isn't a superset of SNPs (I am a bioinformatics PhD student so while this isn't precisely my day job I'm not speaking from a position of ignorance).

[jessriedel]

Thanks for the correction.

[nutmeg]

Is there an easy way (or introductory guide) for someone not familiar with bioinformatics tools?

[citricsquid]

http://www.snpedia.com/index.php/Promethease

[heuermh]

I've written a free/libre open source java client for the 23andMe API here

https://github.com/heuermh/personal-genome-client

Let me know if you would like any help with the analysis.