Learning Large-scale Bayesian Networks From Pairwise-based Protein-Protein-Interaction Network

The motivation for this paper is to apply Bayesian structure learning using Model Averaging in large-scale networks. Currently, Bayesian model averaging algorithm is applicable to networks with only tens of variables, restrained by its super-exponential complexity. We present a novel framework, called LAMA (Large-Scale Model Av-eraging), making it possible to handle networks with very large scale by divide-and-conquer.The method of LAMA comprises three steps. In general, LAMA first perform-s the partition by using a second-order partition strategy, which achieves more robust results. LAMA conducts sampling and structure learning within each overlapping com-munity after the community is isolated from other variables by Markov Blanket. Finally LAMA employs an efficient algorithm, to merge structures of overlapping communities into a whole.In comparison with other four state-of-art large-scale network structure learning algorithms such as ARACNE, PC, Greedy Search and MMHC, LAMA shows compa-rable results in five common benchmark datasets, evaluated by precision, recall and f-score. What’s more, LAMA makes it possible to learn large-scale Bayesian structure by Model Averaging which used to be intractable.Finally, we extend LAMA to LAGE. LAGE is a systematic framework developed in Java. The motivation of LAGE is to provide a scalable and parallel solution to re-construct Gene Regulatory Networks (GRNs) from continuous gene expression data for very large amount of genes.