The research came about from two problems we wanted to explore, namely what is the universe of known drug target space and how could one rationally triage a genome for drug targets. To our surprise, at the time we initiated this work there was no comprehensive survey of known drug targets and their chemical leads. It is an interesting observation of the culture of drug research that, while everyone is working on the "next big thing," there has been surprising little historical understanding of the chemical-biological space covered by drug and their targets. Understanding drug target space is essential in attempting to integrate bio-informatics and chemo-informatics. The size of the "druggable genome"—the identifiable universe for pharmaceutical interventions—is an important strategic question. During the last decade of the twentieth century, one of the underlying assumptions in pharmaceutical strategy was that there would be too many targets from the human genome for the industry to handle. The results of our analysis challenge that underlying assumption.

The paper outlines how the opportunity to find new drug targets for traditional orally available drugs, may be significantly smaller that previously assumed. The chance of finding a new drug is fundamentally limited by the physical nature of the drug target—the target’s "druggability." Increasing our understanding of the properties of drugs can lead us to understand the complementary properties of what makes a good drug target. Combining structural biology with chemo-informatics and bioinformatics can help the estimate of how many drug targets there may be coded within the human genome. These potential targets are defined as the "druggable genome." Analysis of the genome forming a "first principle" approach to "what makes a good drug target" suggest that about only 10%-14% of the genome may be identified as potentially tractable with small molecule drugs. Drug targets not only have to be "druggable" but have to be clinically effective as well. Results from the large-scale mouse gene knock-out studies independently suggest that only about 10% of genes may actually exhibit clinically exploitable phenotypes. Since "druggablity" and clinical efficacy are independent parameters, the overlap could be as small as only as a few percent of all genes in the human genome—an unexpected and controversial conclusion. The size of the "druggable genome" has important implications for pharmaceutical research. Directing research towards chemically tractable targets, which are more likely to deliver clinical candidates, can have major implications for increasing drug research productivity.