Reverse engineering of regulatory relations in gene networks by a probabilistic approach

In the last years microarray technology has revolutionised the fields of genetics, biotechnology and drug discovery. Due to its high parallelity, different analyses can be accomplished in one single experiment to generate vast amounts of data. In this paper we propose a new approach to solve the reverse engineering of regulatory relations task into gene networks from high-throughput data. We develop an Inference of Regulatory Interaction Schema (IRIS) algorithm that uses an iterative method to map gene expression profile values (steady-state and time-course) into discrete states, so that, a probabilistic approach can be used to infer gene interaction rules. IRIS provides two different descriptions of each regulatory relation: the description in which interactions are described as conditional probability tables (CPT-like) and descriptions in which regulations are truth tables (TT-like). We test IRIS on two synthetic networks and on real biological data showing its accuracy and efficiency.