Regulators and insurers have already told us what we need for provide for PGS devices, but most people are not aware of it. What is necessary are at least these two things:
- Analytical Validity
- Clinical Validity
23andMe was dinged for not having either of these two pieces of evidence. Analytical validity evaluates the precision and accuracy of your assay AND your informatics tools. Clinical validity is showing that the marker actually distinguishes, says, BRCA positive patients from negative.
I believe every product on the market should have verified analytical validity that goes BEYOND what CLIA expects. Clinical validity for something like PGS is harder to do. Right now regulators are depending upon evidence drawn from drug development and expecting to see the same kind of thing for PGS devices. But this isn't realistic as PGS offers different kind of, clinically relevant, information. But it's not up to bioinformaticians to decide upon clinical validity either, in fact we need to co-operate with statisticians to propose reasonable studies and level of evidence of these new types of devices - this is starting, but not a big movement.
If you want to know what FDA acceptable evidence is, look at OncotypeDx Breast articles, they have all this levels of evidence the FDA expects, and also check out these PLoS collections: http://currents.plos.org/genomictests/. The scope of these examples still isn't the same as a PGS like what 23andMe offers, but again, this is what the FDA understands.