Physician-geneticist Francis Collins has gained worldwide renown for his leadership of the Human Genome Project (HGP), as well as for his landmark discoveries of disease-related genes, including the gene for cystic fibrosis. While at the HGP’s helm, he worked to be sure that genomic information would be free and accessible to the global scientific community. Collins later served as project manager for the International HapMap Project, which constructed a freely available catalogue of human genetic variations and their chromosomal organization.

Since 1993, he has directed the National Human Genome Research Institute (NHGRI), one of 27 institutes and centers that make up the National Institutes of Health (NIH) in Bethesda, MD. From his unique vantage point, Collins sees the potential for rapid advances toward individualized, prevention-oriented medicine.

He discusses his own laboratory’s current work with type 2 diabetes, the potential of “big science,” and the possible trajectory of personalized medicine as it will unfold in the future.

What We Never Dreamt We Could Do

Q: Drawing mainly from your lab's emphasis on diabetes, how are scientists today utilizing the information from the HGP and the HapMap to find the causes of complex diseases and reach toward treatment?

A: This is an enormously exciting time for unraveling those mysteries. My lab has worked on the problem of type 2 diabetes for more than ten years. We made some steady but slow progress over that decade. Now, we are exhilarated to engineer a leap forward that is unlike anything we have dreamed of being able to do.

Q: What is that “leap?”

A: We have just carried out an unprecedented genetic analysis of 1,300 Finnish people who have diabetes. We know a lot about every one of those individuals, because of their dedication to the project, the Finnish medical record system, and general enthusiasm for medical research in Finland. We have also studied an equal number of non-diabetic controls. In the span of about three months, the Center for Inherited Disease Research has managed to conduct approximately 317,000 different genetic analyses for each individual.

Q: 317,000 for each of the 2,600 people?

A: Yes. That’s a pretty impressive amount of work, an astonishing leap forward, compared to what we were able to do as recently as a couple of years ago.

Q: What made this breakthrough possible?

A: The HapMap, and advances in genotyping technology. These two events have really ushered in a very new era in terms of the ability to unravel hereditary factors in complex disorders like diabetes. We know that each of us has a genome that is 99.9% the same as everybody else. Thus, we are an awful lot alike at the DNA level. But that 0.1% of DNA that is variable, it carries the reasons why I might be at risk for diabetes and somebody else might be at risk for cancer.

If you look at a large population of people, there are ten million places in the genome where there are these common variations. Until HapMap came along, you would have to test all ten million variations in large numbers of affected and nonaffected individuals to begin to understand the causes of disease.

Q: So, in your group of 2,600, you'd have to test all of those 10 million in every single person.

A: That’s right. But in our study, we didn't have to analyze 10 million, only 317,000.

Q: How did you manage that shortcut?

A: The HapMap characterized the fact that these variations in DNA sequence, which we call “single nucleotide polymorphisms” (SNPs), are not all independent of each other. Rather, they travel together. You can imagine them as “neighborhoods” of variations.

Q: That’s an interesting metaphor. What does it mean?

A: Genetic variation is organized on chromosomes in “neighborhoods.” Within the neighborhood, all of the variation is tightly correlated. We learned that if we know the boundaries of those neighborhoods, then we could merely pick out two or three SNPs to represent the whole. In other words, a smaller set of SNPs could basically serve as proxies for the entire neighborhood.

Q: How is that related to the HapMap?

A: The HapMap allows us to pick the optimum set. Then, with fewer SNPs, we can cover the whole territory. The HapMap is a shortcut and saves us, by about a factor of 30, the amount of work that we have to do. The other critical development that has made this kind of project possible is a dramatic drop in cost of testing one SNP on one DNA sample, which has fallen from about 50 cents four years ago to about one-third of a penny today.

The Big Science Revolution

Q: This success gets to the issue of how “big science,” meaning large-scale multi-team projects, have revolutionized both today's laboratory practices and medicine of the future.

A: I remember in 1989, when my lab, working with collaborators in Toronto, was finally successful in identifying the cystic fibrosis gene. I remember being called by somebody at the NIH saying, “We want you to come to a workshop to tell us how you would apply that same strategy for diabetes.”

I said, “It's never going to work. It's too hard.” The idea of being able to do that kind of a survey of the whole genome for diabetes susceptibility genes seemed so complex; I could not imagine it, in 1989, as something that could be done in my lifetime.

Q: And yet, you proved yourself wrong.

A: Yes. Right now, here we are looking at our data on diabetes, and comparing them to similar analyses from other groups in Boston and the United Kingdom. Though it is early in this new phase, we are seeing a number of interesting genetic variants emerge that could shed fascinating new light on the origins of this disease.

Q: So, given this success, is biomedical research moving toward these more large-scale government-funded programs, in tune with the HGP or HapMap Project?

A: Yes and no. There are indeed some exciting opportunities for large-scale genetic analysis of human illness. We recognized that with the HapMap tool in hand, we had to come up with some way of speeding its ability to find the causes of common diseases such as diabetes. Thus, there are several major initiatives to do that (see sidebar). But most of the major insights into the causes and potential cures of disease will continue to be made by individual investigators, empowered by these new genomic tools and the free access to large data sets.

The Ripple Effect

Q: Because of this shift toward larger projects and the complexity of data emerging from them, do you see a parallel change in the sociology of science?

A: Absolutely. Traditionally, release of scientific data occurs at the time of publication. But for many large, complicated, long-term studies, that may take a long time. Still, many people would benefit from getting early access to such “community research projects,” where the real goal of the project is to provide a resource that many investigators can use. Thus, the ethic is really shifting in the direction of making data available for such projects long before the time of publication, so that individuals who have good ideas about how to use it can get started.

Q: How should the research community juggle the sometimes competing interests of the academic, government, and private sectors in order to realize the promise farther downstream?

A: We should make sure that these community resource projects get done quickly and efficiently, but also with maximum attention to immediate data access. That balance will help people in all three sectors to build on this pre-competitive data to make the contributions they want to make.

A Trajectory toward Personalized Medicine

Q: Moving from the bench toward the clinic, are we realizing the promise of research breakthroughs in medicine, or is the effort still largely concentrated on discovery in laboratories?

A: We are seeing the leading edge of it now, but I think the full flowering lies ahead of us.

Q: How do you think that will happen?

A: I see the output of this basic research discovery revolution affecting clinical medicine in three ways:

One, the ability to make predictions about who is at high risk for disease while they are still healthy. We are doing that already in some well-defined instances such as with people who have strong family history of colon cancer or breast cancer. But for most diseases, such as diabetes, we are not quite there yet.

Two, pharmacogenomics, surveying people’s genomes for variations in order to identify which are associated with beneficial or harmful drug responses.

Three, therapeutics. To me, as a physician, that's the most exciting. It's also the one that has the most steps involved. It’s one thing to identify a promising drug target, and quite another thing to have an FDA-approved drug that works, based on that knowledge. But it will happen. The pharmaceutical industry is pretty much unanimous that the one of the most promising pathways toward the identification of validated drug targets is basically the study of human genetics, the kind of variation studies that we are now doing for diabetes.

Q: Following your trajectory, one, two, and three, is that the way you think personalized medicine might be implemented in the next five to ten years?

A: I think the diagnostic and pharmacogenomic applications will come along sooner than the therapeutics. But, there will be exceptions. Every disease is going to travel this pathway on a different pace.

Q: Are there obstacles to your predicated course?

A: Yes, there are many. One obstacle is that we are lacking in this country a long-term, large prospective national sample upon which to conduct genetic studies.

Q: Would there be privacy issues with such a prospective national sample?

A: Absolutely. Another obstacle is that we still do not have effective federal legislation in this country to prevent the use of genetic information to discriminate against people in health insurance or the workplace.

Looking Ahead

Q: What do you see for the more immediate future?

A: I would predict, in the next two or three years, that we are going to see an absolute explosion of information about the genetics of common disease. In fact, you can already see the leading edge of that with recent publications on macular degeneration, on prostate cancer, on Crohn’s disease, and on type 2 diabetes, with many others not far behind.