From Promoter to Expression ΓÇô A Probabilistic Framework for Inferring Regulatory Mechanisms

Опубликовано 6 сентября 2016, 5:17

Inferring regulatory mechanisms based on in silico analysis of regulatory elements has been the target of much research efforts in recent years. Specific aims include identifying combinatorial interactions of either known or novel transcription factors; identifying the cellular conditions in which these combinations are activated and which genes they control; pinpointing the active binding sites of each transcription factor; and finally predicting the expression profile of a gene given its promoter sequence. I will present a sequence of works that use a flexible probabilistic framework to handle these tasks. The framework we developed is based on probabilistic graphical models, such as Bayesian Networks. Using these models we aim to combine diverse sources of genomic data in a synergistic manner and then use the learned models to generate biological hypothesis. I will demonstrate the possible gains from these models when analyzing transcriptions factors from the TRANSFAC data base and when analyzing the Yeast genome combined with various high throughput data.