Prof. Simon Kasif (Boston University and Children's Hospital, Boston)
Thursday, 10.7.2008, 13:30
We describe several new methodologies for developing and evaluating novel diagnostics for diabetes and cancer. In the first part of the talk we describe the Gene Network Enrichment Analysis (GNEA) methodology for identifying molecular markers for diabetes. This approach suggests that network based techniques might be more sensitive and accurate at identifying disease progression than single genes. At the same time we should pay significant attention to both the biological interpretation of results and the biases introduced by network analysis.
In the second part of the talk we provide a perspective on cancer biomarkers using rigorous comparative evaluation of clinical and genomic biomarkers. We evaluate a number of genomic biomarkers inspired by biological, clinical and statistical motivation and derive surprising results on their relative utility and accuracy.
Taken together these results suggest the significance of close collaborations between biologists, clinicians and computational researchers in order to properly evaluate the utility of genomic and computational methodologies in the context of clinical deployment.
This work is joint research performed at Boston University, Harvard Medical
School, Joslin Diabetes Center, Harvard School of Public Health and the
National Center for Biomedical Computing (I2B2) at Harvard Partners.
Host: Prof. Ron Pinter