I would like to design a SNP chip to be used in primary care/clinical setting. I have the ability to determine what markers to have on this custom designed chip. This is not going to be focused on a single disease group/phenotype and is meant to be used as a screening test, risk prediction tool and for some known diseases a diagnostic test. I can think of looking up OMIM, ClinVar and HGMD Professional for known disease associated SNPs. Do you have any other suggestions? What should be the best approach for designing a chip like this?
This is a pretty tough question that I think is less about the nuts and bolts of bioinformatics and more about problems designing diagnostic tools and using "high-tech" genomics in a clinical setting.
I feel like the best place to start is in defining what is encompassed in "primary care/clinical setting". Who is the intended user and what are the things that will be useful to them? There is no shortage of SNP data, but SNPs associated with very rare diseases or for diseases where the SNP has limited or no value in a clinical context might not be worth having. I guess an easy place to start would be including SNPs for which there are FDA approved assays for, from there you could expand.
I guess the other question is how much do you want this to be a "here and now" tool versus having the potential to aid in future research. If you want it to be usable for research, it might be worth including SNPs that don't have much clinical use due to limited insight on how that variant impacts disease/treatment. Having this kind of data on tons of patients could allow for some interesting analysis.
The other problem is that if you want it to be something that could be used outside of a clinical research setting, you'd have to get FDA approval. The word "diagnostic" is a pretty serious thing. This is where 23andMe got in a ton trouble. Obviously not an issue now, but it is something to think about.
I really feel like you'd have to sit down with some physicians and ask them as well as someone with an understanding of how the FDA approaches tools like this.
Thanks for the thoughtful response. The chip in question is designed to be used in a primary care setting. Meaning that it is intended to serve as a diagnostic tool for known disease associated snps (may not be the ultimate tool but a diagnostic tool that can point the lab in the right direction which then will be validated) and the idea is to create a CLIA validated chip. We should also have pharmacogenomic variants that affect drug response or tolerance. It should also give some disease risk information similar to what 23 and me chip does. I am talking to physicians and clinicians but no clinician has the "forest view" when it comes to designing a chip like this, I feel it is ultimately an informatics task to decide what individual snps to be on here as the criteria for 750.000 snps can only be applied in an automated way.