The Precision Life presentation I found interesting. I think it was stuff we have seen before but the presenter was clear and I got a bit more of an idea what it was about.
I think they have been doing roughly what I hoped Chris would set up when I first saw the results - looking for which risk-associated SNPs clustered together, if they did. The rationale Precision Life use is I think misguided. They find clusters of SNP associations and assume that these tell them what the disease pathways are involved in those cases. I don't think it will work like that. A disease mechanism will involve tens of thousands of gene products along the way and any one of those could tip it one way or another. Risk genes are likely to be more central to pathways but you cannot say that Ms X with SNP G has a particularly G involving disease. The influence of the genes is statistical and for most of those in DecodeME fairly slight. The main reason why Ms X and not Ms Z got ME/CFS is probably entirely random - something everyone seems to forget.
Nevertheless, the method of looking for clustered SNPs might tell us something very useful about how risks interact or don't. What I would like to see is an analysis of their sort just on the 8 main regions for DecodeME. They include a whole lot of other genes that presumably have lower level signals. They may be relevant but I want to know, for instance, whether OLFM4 clusters with BTN2A2 and not RABGAP or whatever. I guess they have those data.