Nicholas Wade (guesting posting for John Tierney) points to a set of papers in this week's Nature reporting large genome-wide association studies of schizophrenia. The main upshot of the papers is that variants in the MHC region influences susceptibility to schizophrenia. This region influences susceptibility to a number of autoimmune diseases, so the association is suggestive evidence that schizophrenia as well has an autoimmune component.

Outside of the MHC, however, there are few convincing signals of association. One interpretation of this might be that there are simply no other common polymorphisms that influence risk of developing schizophrenia. One of the groups, however, set out to test whether this was the case. They took thousands of the top associated SNPs--none of them individually showing a strong association with the disease--and assembled them into a genotypic score for predicting whether an individual has schizophrenia. And indeed, using these thousands of markers, they were able to do significantly (in the statistical sense, not really in the practical sense) better than random at classifying individuals as schizophrenic or not from genotype data alone. Thus, the aggregate effect of thousands of polymorphisms impacts the development of this disease.

What is the practical significance of this? In terms of treatment or drug development, there is essentially none. But it does suggest that there will be no "silver bullet"--copy number polymorphism, rare variants, or what have you--that will solve schizophrenia genetics.

Labels: Genetics