Genomics, of course, will transform traditional empirical medicine into rational treatments for specific pathologies. So with two working drafts of the human genome now officially published, what can we realistically expect from genomic medicine?

The theory goes something like this: Find the genes involved in a disease and identify the encoded proteins. Accumulate information on protein structure and function, elucidate biochemical cascades, and select key control proteins as potential drug targets. Identify compounds that interact with the target and use structural information to refine binding affinity and in vitro data/in silico prediction to optimize toxicological/pharmacological properties.

That entire process should take about 10 to 15 years (with a following wind). The first step of the approach—identifying novel drug targets—has progressed. Searches of the genome sequence have already revealed at least 30 new disease genes (see p. 207). But as everybody is keen to say these days, this is only a beginning.

Large gaps exist in our understanding of the processes that influence protein diversity and function. Current estimates suggest that at least 10,000 human genes undergo splicing (see p. 196), yet we still know little of this process. Many disease genes are expressed at low levels, and gene expression often shows little or no correlation with changes in the levels of protein anyway.

All this means that we urgently need sensitive and reliable proteo-mic methods for identifying proteins on a large scale and characterizing post-translational modifications that influence function. Modeling of genetic networks and metabolic engineering also are rudimentary, making elucidation of key points of therapeutic intervention difficult.

There are also many layers of complexity that drug development has to address. There may be hundreds of different variants associated with a single predisposing gene, complicating the design of a single small molecule. How much worse will this problem be for multigenic disorders? Validated protein targets, when found, may not crystallize to allow their structures to be solved. The key proteins of many diseases (e.g., sickle cell anemia) have been known for years, without leading to effective drug treatments. Even when small molecules bind to target proteins, that simple interaction may not reverse the complex perturbations of biological networks that occur in disease.

Most traditional pharmaceuticals combat disease by antagonizing drug targets, thereby ameliorating gain-of-function mutations. Targets uncovered by genomics, on the other hand, are more likely to result in loss-of-function mutations. Addressing those targets will require novel agonist drugs with new chemistries that confer appropriate binding site kinetics and distribution within the body. A recent report “The Fruits of Genomics” from Lehman Brothers estimates that the number of drug candidates entering the clinic in the pharma industry will increase by fourfold in the next five years. The productivity of the drug discovery process may, therefore, increase significantly. But the newness of the targets and of the chemistries used for drug candidates will mean significantly higher rates of clinical failure. Thus, genomics will not rapidly improve the efficiency of drug development. In fact, it may make it even more complicated.