Is the Era of Personal Genomics Over?

An article reported in the New York Times this month highlighted concerns of geneticists and physicians over the use of personal genomics as a method for predicting diseases. Many companies have sprung up, such as 23andMe ($1000 a test), that use genome wide association studies (those which compare the genomes of patients and healthy people to look for certain alleles that occur in higher frequencies in people who have the disease than in healthy populations) to "assess" a person's risk for a particular disease. The companies take a DNA sample sent to them by the customer and sequence particular loci to look for allele variants that have been shown to occur in a higher frequency in people with the disease than in healthy individuals. Recently there has been outstanding criticism over the value of these association studies in predicting diseases. It seems that they reveal little about the genetic link to most diseases relative to the multimillion dollars they cost to produce. Do geneticists continue on this path of looking for common alleles found in the diseased population (by common, this usually means 4% of those who have the disease have this allele, versus 2% of healthy individuals have the allele, thus the person is theoretically twice as likely to contract the disease) or do they take a new direction and start sequencing the entire genomes of individual patients. The problem lies in that the diseases being looked at such as cancer, diabetes, heart disease are not the result of one or a few allele variants with pleitropic effects, which is the case for most rare genetic diseases suc as cystic fibrosis. Our assumption in genome wide association studies is that since these diseases are common in the population, we expected to find common genetic variations that cause these diseases in diseased populations but are absent in healthy populations. In taking this approach so far, we have found a few variants that are associated with some common diseases but they only a moderate disease risk at most has been shown. Some scientists argue that the confounding factors are large numbers of rare varients that will be undetected by performing genome wide association studies (a contradiction to those who believe that the common variants will identify the biological pathways through which a disease emerges, leading to possible drug targets). The possibility that hundreds of rare varients interacting to create a disease such as cancer makes a drug targeting a specific biological pathway seem daunting, if not impossible. Genome wide association studies will fail to find the rare variants with small effects and our resources should be put towards decoding full genomes of select patients. Others argue that genome wide association studies are useful tools in predicting risk for certain diseases, especially if one only considers diseases in which one's genes predict a 3 fold or greater risk for the disease than normal. Some scientists maintain that the underlying factor for risk is not rare allele variants, but unexpected biological mechanisms. It is unlikely that the era of personal genomics using genome wide association studies is over, however, sequencing entire genomes to discover rare variants is a worthy use of our time and resources.