Marco Ramoni, PhD
Instructor, Harvard Medical School
Research Associate, Children's Hospital
The massively parallel data generation capabilities of microarrays enable the acquisition of snapshots of the transcriptome. These snapshots have been analyzed using different clustering methods in order to characterize genes' functional behaviors. Mounting these snapshots into movies, portraying the temporal evolution of the transcriptome under certain conditions, is a fundamental step to dissect and understand the global behavior of a cell. This talk presents a Bayesian clustering method for temporal profiling of gene expression data. The method searches for the most probable set of clusters, conditional on the data available, using an agglomerative procedure and a simple model description of the temporal evolution of gene expression data. The method identifies, automatically, the most probable number of clusters given the data available, and provides a simple model description of the temporal evolution of gene expression data in each cluster. The talk will also include a demonstration of CAGED (Cluster Analysis of Gene Expression Dynamics), a computer program implementing the theory and methods described in the talk.
CAGED website: http://www.chip.org/caged