The Colorectal Cancer Serum Proteomics Bakeoff

Daniel Normolle, Ph.D.
University of Michigan

Wednesday, February 14, 2007
3:30pm
Moos 2-690
Minneapolis Campus

Abstract:

Proteomic profiling attempts to leapfrog over difficult cellular process elucidation and screen a population for cancer by employing a scalable platform, such as MALDI-TOF-MS, to extract a "fingerprint" of a bodily effusion, such as serum, and then training a statistical classifier to recognize fingerprints associated with cancer. Several laboratories have claimed success, but the validation of these claims has proved elusive. The Great Lakes-New England Clinical Epidemiology Center of the NCI-funded Early Detection Research Network implemented the Colorectal Cancer Serum Proteomics Bakeoff to attempt to validate the proof of principle of this technology. Dr. Normolle will describe the design of the Bakeoff, discuss issues statistical and scientific surrounding its conduct, show how he used SELDI-TOF and MALDI-TOF-MS spectra generated by the Eastern Virginia Medical School to separate cases and controls, and reveal how he fared compared to groups at Johns Hopkins University and the University of Alabama at Birmingham. Along the way, he will briefly encounter wavelets, Cromwell, neural nets, support vector machines and bootstraps.

A social tea will be held at 3:00 P.M. in A434 Mayo. All are Welcome.
For more details contact 612-624-4655 or see http://www.biostat.umn.edu/seminar_academic.html