Abstract:

Time series gene expression experiments are becoming a popular method to study biological systems. The time series data produced by these experiments raises a number of computational and statistical challenges. In this talk I will first survey the work of Bar-Joseph et al. that develops continuous representation of time series gene expression data and then uses the representation to cluster genes and identify differentially expressed genes. I will then describe recent research that I have done to identify statistically significant sets of co-expressed genes in short time series data. I will close by describing new directions that I'm exploring to fuse time-series gene expression data with static data sources to infer physical network models of gene interactions. No prior biological knowledge will be assumed. Work described is joint with Ziv Bar-Joseph and Srinath Sridhar.