Explaining Genetic Knock-out Effects Using Cost-Based Abduction
Emad Andrews and Anthony Bonner
Cost-Based Abduction (CBA) is an important AI model for reasoning under uncertainty. In CBA, evidences to be explained are treated as a goal which is necessarily true and to be proven. Each feasible explanation is a proof for the goal and has a cost equal to the sum of costs of all hypotheses that need to be assumed to complete the proof. The goal is to find the Least Cost Proof. This paper introduces the first application of CBA in Computational Biology (CB). We introduce a novel method to model Genetic Regulatory Networks (GRN) and explain genetic knock-out effects using CBA. Constructing GRN using multiple data sources is a fundamental problem in CB. We show that CBA is a very powerful formalism that can easily and effectively integrate multi-source biological data to model GRN. We use three different biological data sources: Protein-DNA, Protein–Protein and knock-out experiments data. An un-annotated graph is created using these data sources; then, CBA annotates the graph by assigning signs and directions to all edges. Our biological results are very promising; however, this manuscript focuses on the mathematical modeling of the application. Relation to Bayesian Inference and advantages of using CBA are also presented.