Colloquium: Prabhani Don

Speaker: Prabhani Don, Dana-Farber Cancer Institute, Boston, and Harvard T. H. Chan School of Public Health, Boston

Title: Finding Hidden Patterns in Genomic Data: Recent Developments

Abstract: The first part of the talk focuses on discrete latent variable models that allow one to characterize the structure of two-way data arrays. On large datasets, estimating these models by maximum likelihood is computationally unfeasible. We thus develop and employ composite likelihood modifications for our models, and use them as a computationally tractable alternative to the full likelihood. The performance of our proposed methods is illustrated via simulations and applications to genomic data.

The second part of the talk focuses on open methodological challenges related to RNA expression profiling in formalin-fixed paraffin-embedded (FFPE) tissue samples. Over 20 million such samples archived each year in the US alone represent an enormous potential resource for the genomic study of cancer - which remains under-utilized to date due to the difficulty of appropriately generating and analyzing expression information.