Gene network inference via sparse structural equation modeling with genetic perturbations

Structural equation models (SEMs) have been recently proposed to infer gene regulatory network using gene expression data and genetic perturbations. However, lack of efficient inference method for SEMs prevents practical use of SEMs in the inference of relatively large gene networks. In this paper, relying on the sparsity of gene networks, we develop an efficient SEM-based method for inferring gene networks using both gene expression and expression quantitative trait locus (eQTL) data. Simulated tests demonstrate that the novel method significantly outperform state-of-the-art methods in the field.