I agree it's time to stop wasting monies on "genome-wide association studies" (GWAS) and also on 'massively massively parallel" sequencing.
Michael Lerman, Ph.D., M.D.

The identification of disease-causing genes is only the first step in a long road towards successful diagnosis and treatment of illness. Take sickle cell anemia as an example. The common mutation that produces this disease was originally reported by Vernon Ingram et al. back in 1956, after Linus Pauling and his colleagues linked hemoglobin defects to sickle cell anemia earlier in 1949. Today, despite decades of ongoing research, the average life expectancy for victims of this disease is still between 53 to 60 years in North America and Europe.

Recent reports, including findings of the occurrence of typically around a hundred loss-of-function gene mutations in the average healthy human genome (Daniel MacArthur et al.), and the lack of increased risk for 24 different diseases seen when 53,666 identical twins were compared to the general population for the same diseases (Bert Voglestein et al.), exemplify the limitations of disease risk assessment based on purely genomic sequence information. Environmental factors clearly have a major influence on whether most common diseases will materialize, and these can exacerbate or compensate for genetic defects.

Recent genome sequencing studies have yielded hundreds of possible leads for the study of oncology alone. However, most of the major oncogenes were actually identified almost 20 years ago. While thousands of gene microarray studies have been performed to examine changes in gene expression with cancer, it now turns out that gene co-expression data rarely identify direct physical linkage between the protein products of these mRNA. I noticed at the most recent American Association for Cancer Research Meeting in Chicago, an increasing emphasis on epigenomics and micro-RNA research. While interesting from a basic research perspective and this may even have some diagnostic value, these directions are unlikely to translate into improved therapies for most cancers and other diseases. The phenotype for almost any disease ultimately depends on the complex interplay of normal and defective proteins, and for host of practical reasons, proteins remain the best drug targets.

Clearly a myriad of extra levels of control in addition to the genome DNA sequence are exerted inside of cells to maintain their healthy state. These include epigenetic, post-mRNA transcription, post-protein translation, and allosteric regulation. Further complexity arises from hormonal and nervous communications between cells. Our growth in understanding of these processes from biomedical research over the last 50 years has been impressive, but our knowledge is still extremely fragmentary. I remain constantly frustrated by the relatively little support afforded by government and charitable organizations towards looking beyond DNA and RNA. As long as proteomics and metabolomics research continues to be short-shrifted, there will always be a translational gap, and the general public and politicians will become increasingly dubious about their biomedical research investment.

Realism sets in but this is far from despair. WGS is still the best for families with many affected members for both rare and complex disease. Please don't forget that whole genome sequencing is roughly as cheap now as arrays were when Affy introduced its first whole genome microarrays. With arrays, GWAS is very cheap now (c.110 dollars for reagents per sample for large scale studies) whereas measuring 5-6 hormones levels costs that much per sample. Yes you find only a few real hits per experiment but those are, with replication built in, real hits which tell us a lot about both the genetic architecture and the biology of the disease studied. Sequencing will become just as cheap and will probably tell us more, within a decade at the latest.