Bayesian model for genome-wide prediction of tissue-specific gene expression

Wspólne seminarium badawcze IPPT i MIMUW.

I will discuss a new probabilistic approach to predict tissue specific expression genome wide.  We use a Bayesian approach to find model parameters maximizing the likelihood based on genome wide data on Transcription Factor binding and gene expression. The model takes into account 3 different levels: transcription factors, enhancers and
promoters. We will show the resulting model is capable of accurate predictions of new tissue-specifically expressed genes in Drosophila mesoderm development.