Now that the sequence of the human genome is virtually complete, what is the single most pressing scientific problem you would like to solve with the information it provides?

That’s very clear. I got excited 15 years ago about the genetic basis of human variation; polygenic traits, and about genes that cause disease. In 1985 or '86, David Botstein and I began working on methods that, in principle, would let you dissect complex genetic traits in natural populations like humans. We realized, however, that you would need extremely dense maps of genomes to do this. You would need all sorts of such tools. For me the genome project was very much a means to an end. In effect I took a 14-year digression to get this set of tools, so I could then figure out the nature and identity of the genetic variance responsible for genetic components in populations. That's why we also put together the SNP [single nucleotide polymorphism] Consortium. Three years ago we wrote a paper on the large-scale identification of SNPs. At the time we only had 4,000 SNPs, but it lit a fire under the idea of collecting SNP variations. Now we have one and half million in the genome. The totals are growing rapidly. We're getting close to having a fairly comprehensive catalog of all common human variations. Then we can begin to start correlating that with disease, and also working out the structure of ancestral chunks or segments in the human chromosomes. We're trying to identify all variations in the genome and figure out how those variants are correlated with each other and how they're correlated with disease. It's a very clear program. I know exactly what I want to know.

How has the estimate of only 30,000 to 40,000 genes in the genome changed the program for genomics research?

It's made it much easier. We had had a pretty good handle already on about 14,000 genes. The prospect of there being 100,000 genes meant we actually had full-length transcripts or decent descriptions of only about 14% of the genome. Now it means we have that for 40 or 50%. That’s wonderful. Needless to say, you can make the case that if there are only 30,000 genes, it has to be more complicated than if there are 100,000 genes. But it's still a good thing. Fewer genes means it will be even sooner before we have pretty comprehensive descriptions of where the genes are expressed and under what circumstances. It means the periodic table, in effect, is three times smaller.

How do you take the information contained in the genome and turn it into pharmaceuticals?

You need first to have a reasonable idea of the  targets. In the past we were target-limited. Now we have a huge abundance of targets, but not much annotation of which targets are good targets. The challenge is to identify the good targets. That requires correlating genes with disease, which could be done by human genetics or by expression studies or by having ways to inactivate a gene and see how it affects the phenotype of an organism. Having the genome means we can begin to take global views. We can ask,  in a particular type of cancer, what genes are always turned on? In the past we were hunting and pecking. Now we can get the best candidates that emerge from comprehensive analysis of all genes. We have to come up with predictors of how well inhibiting a certain protein will affect disease and how likely it is that a particular molecule will cause toxicity or be metabolized. We need to raise the probability of success in  drug development. We have to turn drug development into engineering. The genome gives the components. Then we have to get the interactions. Genomics is the first step along the way.

What does it entail to make biology predictive? To turn it, in your words, into engineering?

It means complementing this component-by-component bottom-up approach with a global top-down approach. It's easier said than done. The top-down approach requires a way to parse the complexity of cells and tissues into meaningful modules—circuits, in effect—and we don't know what we mean by this. We talk about circuitry and systems but we don't have a very good feel for what these are.

What technologies are really going to make the difference in achieving this?

It's not any one technology. It really is the ability to simultaneously look at DNA variations, RNA levels, protein levels, and small-molecule interactions. There is a tremendous explosion of genomic information that we can now get, and no one piece is going to tell us that much. We're going to have to deploy all these tools simultaneously. The buzzword lately has been proteomics. The newspapers say, “We did genomics, now we're going to do proteomics.” Well, that's just a subset of genomics. The new genomics is a global point of view in biology. It's going to take us a couple of decades to work it out and interpret that view. It's not going to be affected that much by the fad of the month or of the year. It's going to be a sustained enterprise to understand the global integrative view of biology.

How did you end up first author on the genome paper from the Human Genome Project?

That was easy. By the amount sequenced. The Whitehead was the largest contributor to the human genome project, so we got to be first group listed. That was a great honor. Also a responsibility, because it meant we had the responsibility of organizing the manuscript, which I really enjoyed. It was all-consuming but worth it. I got to hole up in the seventh-floor faculty lounge of the Whitehead for about four months, working together electronically with many other colleagues, on the text.

The paper is written for graduate students, post-docs, and journal clubs, and for lay people who might want to make their way through it. Since Nature was giving us a huge amount of space, we thought, what the hell, instead of boiling the thing down to the usual scientific language, we could take another 20 percent to make it interpretable and understandable. We felt that it was an opportunity to do a tremendous amount of teaching, which distinguishes it from many of the other papers published in science. We really worked hard as teachers to make sure this was a place where the next generation could pick up the paper and say, “Aha, I see what can be done with genomics. I see how this work can be done.” It wasn't literature but it was as close we could get.