Francis Collins is changing the way we think about DNA. First he helped decode the human genome. Now he oversees a research empire at the National Human Genome Research Institute (part of the sprawling National Institutes of Health), leading the race to find disease-causing sequences hidden in our genes, doling out hundreds of millions in grant money each year, and defending controversial science in the charged politics of Washington, D.C.

A devoted Christian, Collins defends evolution and embryonic stem cell research in his new book, The Language of God: A Scientist Presents Evidence for Belief (Free Press, 2006). He dismisses religious extremists and scientist-atheists as equally shrill and believes that both sides push their beliefs on a public who prefers that science and religion remain separate.

In college Collins found biology boring and trained as a chemist, only to become a physician and then a famed gene hunter at the University of Michigan at Ann Arbor, where he identified the gene for cystic fibrosis and helped find the gene for Huntington's disease. He blazed into the public eye in the late 1990s during the bruising competition to sequence the human genome; he ran the $3.7 billion public effort while J. Craig Venter conducted a parallel effort using private funds. After much scientific mudslinging, the two sides declared a tie in 2000 and were feted jointly at the White House.

Moving beyond mere code breaking, Collins has been focusing on the recently completed International HapMap Project, a genomic atlas of clusters of disease-causing snippets of DNA. The project catalogs human variability and identifies patterns of genes that are linked to health and disease.

DISCOVER contributing editor Duncan spent an afternoon with Collins in his offices at the NIH. Collins, a lanky man in well-worn beige jeans and a flannel shirt, has an aw-shucks breeziness about him—an ease that belies his status as one of the most powerful scientists in the world.

You're a born-again Christian who suggests that therapeutic cloning could be acceptable. Some other devout people consider it fundamentally immoral. What do you see differently?
There is a difference between doing research on an embryo that was generated by sperm and egg coming together, which is the way human beings are created, versus the very bizarre laboratory phenomenon of taking a nucleus from a skin cell or the udder cell of a sheep and putting it into an environment that takes it back in time to its stem cell state. In public discourse, they're both called embryos. Even though the somatic cell nuclear transfer approach is a very different biological phenomenon, in many people's minds it has been all blurred together. As a result, we've really missed out on a chance for a much more thoughtful, nuanced discussion, and we're still trying to recover from that.

What kind of reaction has your book provoked?
I didn't know quite what to expect, but the response has been amazing. Most of the large volume of letters and e-mails I have received have been encouraging and positive, both from the scientific community and from the religious community. A few scientists have written that it is inappropriate for a scientist to write about harmony with faith, because they think that faith already has too much power in the United States. A few conservative Christians have been stridently critical about my endorsement of theistic evolution. Most heartwarming have been a few dozen very personal messages from individuals who had been struggling with whether they were forced to make a choice between science and faith and were relieved to hear that it is possible to embrace both.

What did you think you were in for when you signed on to the Human Genome Project in 1993?
When I came here, I don't think that one person out of 10 actually believed that we would sequence the human genome by 2005. We were feeling our way in the dark—we didn't have the methods, the people, the confidence. We were really struggling.

What exactly is a gene?
That's a good question. You ask 100 molecular biologists that and you'll get 110 answers. I have a pretty classic answer—a gene is a well-defined segment of DNA that encodes for a protein. Some genes also code for segments of proteins. The key thing is for a gene to have an exon [a stretch of DNA that transcribes into RNA]. There are also pseudo-genes that encode RNA but have no apparent function. They are holdovers.

Why does the body keep those around?
It's like junk in your basement. Some of it could be thrown out, but some of it you keep around in case some day you need it.

What are you finding about the importance of known genes versus the "junk DNA" in between them?
We're learning a lot about this. At the NIH we have a project called ENCODE—Encyclopedia of DNA Elements. It's a coordinated effort among 30 labs to identify all of the parts of the human genome that have biological activity, including the so-called junk DNA. We have found that there is a lot more action. There are transcription factors happening and RNA being made. We still don't know if all of this activity is actually doing something or if it's the equivalent of background noise, but we're working to find out.

Now that the human genome has been sequenced, what more is there to learn about genetics?
That was just the beginning. Back in 1997, two colleagues and I wrote a paper about the promise of having a really rich catalog of human genetic variation. I thought that was something we should strive for. I didn't think we'd get there this quickly! That project, called HapMap, got started in 2002 with scientists from six countries aiming to lay out how variation is organized across all the human chromosomes in four different populations.

What is a HapMap, and why is it so important?
A haplotype is a stretch of DNA with a particular combination of genetic spellings that vary among different people. Haplotypes in vulnerable gene regions can be responsible for increasing risk of disease. The idea of the HapMap Project was to define the spelling differences in the human population—there are about 10 million of these, called SNPs, or single nucleotide polymorphisms—and how they are organized into haplotype neighborhoods in the genome.

How does that help you learn about disease?
Searching for genetic variations that predict an increased risk of disease can be terribly difficult and expensive. If you had to test all 10 million SNPs in hundreds or thousands of cases and controls for a disease, it would be completely impractical and unaffordable. HapMap provides a valuable shortcut, as it defines the neighborhoods within which the SNP spellings are tightly correlated. In each neighborhood, you can identify a small number of "tag SNPs" that serve as a proxy for all the others that you didn't test. That way, instead of having 10 million things to sample, you can sample using about 300,000. That's what we're doing now, when we look for diabetes genes in my lab, and that's what the whole world is doing with this map for lots of other common diseases.

Has linking genes to diseases become easier since you began your career at the University of Michigan?
Searching for diabetes genes is much harder than anything I did in Michigan. At that time, I was looking for genes for Mendelian conditions, like cystic fibrosis and Huntington's disease: There's one single gene, and if you have the misspelled version of the gene, you are extremely likely to get the disease. Whereas for heart disease or schizophrenia or diabetes, no single gene is going to have a very large effect—these are complex diseases with many genetic and nongenetic causes.

The underlying science is also, frankly, rather tough to explain. Is there a way to describe all of this more simply?
Not really, and it hasn't helped our cause that we geneticists give inscrutable names like "haplotype" to our concepts. This really doesn't help when you're trying to explain this science to the public, or to Congress.