Identification of Causality Among Gene Mutations Through Local Causal Association Rule Discovery

摘要

Detecting the interaction among gene mutations is still an open problem on genetic research. Among various types of interaction, the causality among the gene mutations provides deep insight of the gene mutation and evolution, is the focus of the current research. Different from the global causal network reconstruction method, we propose a local causal discovery method by exploring the causal concept under the association rule discovery framework. Firstly we propose a V-Structure Measure (VSM) to evaluate the causal significance of the local SNPs structures. Secondly, we develop a method called ASymmetric Causal Association Rule Discovery (ASCARD) to mine the reliable causal association rules considering the conflicts among the candidate structures. Finally, the experiments on the synthetic data and WTCCC (Wellcome Trust Case Control Consortium) SNPs dataset shows the effectiveness of the proposed method. Some interesting biological discoveries also show the potential of the real world applications.