The shape of success
Five years from now -- if all goes as planned -- IBM researchers will have computed the three-dimensional shape of a complex protein molecule starting from a knowledge of its amino acid sequence. No one has ever done that.
Such a feat could have revolutionary implications, not only for computing, but for biological science and medicine. In the first instance, however, it will represent the success of a $100 million project that IBM Research announced on December 6, 1999, to build the world's most powerful supercomputer. Nicknamed Blue Gene, the machine will operate at a speed of 1 petaflop (a quadrillion floating-point operations per second), or 500 times faster than today's fastest computers.
Because of the magnitude of the computation, we knew the "grand challenge" problem of protein folding could not be solved by simply scaling up existing supercomputers. Although the proteins in our bodies -- and our bodies are basically protein factories -- fold rapidly and unerringly into the proper shape, the number of possible ways they could fold is astronomical: greater than the number of atoms in the universe. Without knowing what a protein "knows," we must find the sequence of steps through brute-force calculation.
Only a tremendously fast computer could tackle such a problem. Blue Gene will embody a radical advance in architecture, including unprecedented parallelism, a simplified instruction set and a novel "self-healing" design that will allow neighboring chips to take over for any that fail. We know from previous large pioneering efforts that, while we may never offer a commercial system exactly like Blue Gene, what we learn about computer architecture will find its way into mainstream computing, benefiting not only our customers but the entire industry.
Eventually, each of us will profit from the knowledge of how proteins fold. For 50 years, ever since the American chemist Linus Pauling and his collaborators published their pioneering paper on sickle-cell anemia, the prospect of understanding disease at the molecular level has tantalized scientists. Today, with the imminent completion of the Human Genome Project and the new knowledge that Blue Gene will generate, scientists are close to realizing the dream of not just understanding diseases but treating them with drugs tailored to the structure of an individual's proteins.
Blue Gene is the type of bold exploratory project that IBM Research must do. It sets us apart from other research labs and excites the technical communities with which we work. People and organizations outside IBM are already eager to participate, just as we had hoped. So, while we don't yet know if the project will achieve its technical goals, in many ways it has already been a success.
Paul M. Horn
senior vice president and director,
Research
