The field of computational biology will provide researchers with a full slate of important work in the years ahead, but the work definitely can be accomplished, principal investigators concluded after meeting July 21-22 at Cornell to plan the future of their major new collaboration. Computational biology is one of three focus fields of a tri-institutional biological research program announced June 27.
The other two focus fields in the $160 million research program, which is supported, in part, by a major gift from a private donor, are chemical biology and cancer biology. The collaborating institutions are Cornell and its Weill Medical College, Memorial Sloan-Kettering Cancer Center and the Rockefeller University.
The scientists involved in computational biology hope to harness the power of computing to make sense of all the raw data coming from the Human Genome Project as well as other technology-based efforts in biological discovery. Sixteen researchers involved in this aspect of the collaboration met on campus to get acquainted, compare notes and plan the sharing of resources. The meeting was convened by President Hunter Rawlings and moderated by Ron Elber, the Cornell professor of computer science who will lead the computational biology program.
"We're here to talk about the science each institution is doing, establish formal connections and see how we can work together," Elber said. As the scientific show-and-tell proceeded, researchers who had previously been strangers gained an appreciation for the enormousness of each other's tasks.
Epidemiologist Simon Heath from Memorial Sloan-Kettering described the difficulty of discovering all the genes responsible for complex genetic diseases, such as schizophrenia, cancers or heart disease, where each of many factors has a small, contributing effect. "We need (to study) large sample sizes with many markers and large families," said Heath, who uses a statistical technique called Markov Chain Monte Carlo (MCMC) analysis. "The problem in human genetics is that we're not allowed to produce crosses to test hypotheses, as the plant geneticists can. Our work does not require a supercomputer, but it would be nice."
Supercomputers were exactly what Thomas Coleman, director of the Cornell Theory Center, had to offer the researchers. The powerful and constantly updated machines on Cornell's Ithaca campus are particularly adept at a task called continuous optimization, Coleman said, and that has attracted the attention of Wall Street practitioners of computational finance. "We are probably the world's strongest department in optimization," Coleman said, referring to the overlapping departments of computer science and applied mathematics, "and there are many applications (for optimization) in genomics."
By the end of the two-day meeting, the researchers and administrators had agreed to establish a computational biology unit at the Cornell Theory Center with satellites at the New York City institutions.
That was good news for Jurg Ott, who is working at Rockefeller to locate deafness genes. "We're developing microarrays for the purpose, but we're at a loss about how to analyze all the data," said Ott, who also works with the so-called non-Mendelian genes, the multiple factors that underlie complex traits. Just as daunting is another of Ott's projects, to follow several hundred former angioplasty patients in an effort to learn what genetic factors in some cause their arteries to become obstructed again after the procedure.
Rockefeller University's Stephen Burley, a specialist in structural genomics, said his research group is interested in the genetic basis for protein folding.
"Just a few thousand genes are responsible for all possible protein folds," Burley said, "and in five to 10 years, every protein sequence will be within modeling distance. We are very keen on the chemical biology resources at Cornell."
Burley also emphasized speeding research through the Macromolecular Diffraction Facility at the Cornell High-Energy Synchrotron Source (MacCHESS), used by many biologists for X-ray diffraction crystallography. "Utilize robotics to conduct labor-intensive work at CHESS," Burley suggested. A robotic sample-changer, toiling at CHESS -- where synchrotron radiation in the form of X-rays restricts hands-on human involvement -- would make better use of valuable "beam time," the Rockefeller scientist said. It would, he said, "be a good problem" for the engineering school to solve.
Speaking of interdisciplinary problem-solving, Cornell Professor of Operations Research and Industrial Engineering David Shmoys said, "You can't underestimate the power of graduate students in forming such alliances."
The expert in discrete optimization discussed a joint project between his engineering college group and members of Steven Tanksley's laboratory in Department of Plant Breeding to construct genetic linkage maps, noting that the software that results from that collaboration is publicly available on the Web. Tanksley is the Liberty Hyde Bailey Professor of Plant Breeding.
Elber took it as a positive sign that researchers with common interests -- but isolated in laboratories in neighboring institutions -- were "coming out" of their labs and talking about collaborations.
"I spent a year on the second floor (of a Rockefeller University building) and never met the people across the corridor," said Benoit Roux of Weill Cornell Medical College.
Thanks to a convergence of scientific interests and a tri-institutional research program, now he might.
| Cornell Chronicle Front Page | | Table of Contents | | Cornell News Service Home Page |