Moore displays the computational and statistical method that he and his coworkers developed for use in analyzing large assemblages of genetic data, as might be generated by DNA microarray studies.
Moore receives young investigator award
For a genetic researcher to be given an award named for one of the fathers of modern human genetics would be an honor, no doubt. But to say that you had the privilege of being taught by that man, to recall the occasions he dropped by your student office for an impromptu scholarly debate, to remember his kind gaze and keen insight, well that would make the honor special indeed.
For Jason H. Moore, Ph.D., assistant professor of Molecular Physiology and Biophysics, recently receiving the James V. Neel Young Investigator Award by the International Genetic Epidemiology Society was an accolade deeply felt.
“On a personal level, the award means so much to me because I knew him and respected him,” Moore said. “Jim Neel was a great guy with an amazing mind. Into his early 80s, he could run mental circles around anyone. I can only hope I’ll be as sharp at that age.”
The award was given to Moore as one of five nominees chosen to present their research findings at a special session of the 10th Conference of the IGES held last month in Garmisch-Partenkirchen, Germany.
In his oral presentation, Moore discussed the computational and statistical method that he and his coworkers developed for use in analyzing large assemblages of genetic data, as might be generated by DNA microarray studies. The new method, called Symbolic Discriminant Analysis, finds optimal combinations of gene expression variables and mathematical functions that predict clinical outcomes.
Moore described how he and his team used SDA to discriminate between two types of human leukemias, acute myeloid leukemia (AML) and acute lymphocytic leukemia (ALL). Analysis of microarray data measuring more than 7,000 genes, or expression variants, in two independent cohorts of leukemia patients identified a combination of genes and mathematical functions that, Moore said, allowed for “perfect discrimination” between the two types of leukemia.
A powerful supercomputer that Moore helped develop dubbed VAMPIRE, or Vanderbilt Multi-Processor Integrated Research Engine, facilitated analysis of the amassed data by bringing to bear a roomful of parallel processors simultaneously sprinting through the prescribed mathematical and statistical exercises.
“SDA is computationally intensive, since so many combinations have to be tried,” Moore said. “The study would have been impossible without VAMPIRE—it’s been a tremendous asset to my research.”
Most of the genes identified were either directly related to the leukemia process or related to blood cell function or production. One of the genes encodes erythroid beta-spectrin, a major component of red blood cell membranes expressed during normal blood cell synthesis. The adipsin gene is part of a chromosomal cluster of genes expressed during myeloid cell differentiation, and the nucleoporin 98 gene is located at a chromosomal breakpoint that is associated with AML. The CD33 gene encodes a well-known pathological marker of AML.
Statistical analysis of the same data by Eric S. Lander, Ph.D., of the Whitehead Center for Genome Research, previously published in Science, was not as successful as SDA in identifying implicated genes, Moore said.
After the software has been further refined, Moore hopes to offer SDA as a service of the Bioinformatics Core. In the meantime, some of Moore’s collaborators are making use of the method, along with VAMPIRE. Data from the Vanderbilt lung cancer SPORE (Specialized Program of Research Excellence) are currently being analyzed using SDA, and investigators in the department of Rheumatology are using the method to predict the genes involved in rheumatoid arthritis, systemic lupus erythematosis, and other autoimmune diseases.