Gene regulatory network inference using mixed-norms regularized multivariate model with covariance selection

Abstract

Despite extensive research efforts, reconstruction of gene regulatory networks (GRNs) from transcriptomics data remains a pressing challenge in systems biology. While non-linear approaches for reconstruction of GRNs show improved performance over simpler alternatives, we do not yet have understanding if joint modelling of multiple target genes may improve performance, even under linearity assumptions. To address this problem, we propose two novel approaches that cast the GRN reconstruction problem as a blend between regularized multivariate regression and graphical models that combine the L2,1-norm with classical regularization techniques. We used data and networks from the DREAM5 challenge to show that the proposed models provide consistently good performance in comparison to contenders whose performance varies with data sets from simulation and experiments from model unicellular organisms Escherichia coli and Saccharomyces cerevisiae. Since the models formulation facilitates the prediction of master regulators, we also used the resulting findings to identify master regulators over all data sets as well as their plasticity across different environments. Our results demonstrate that the identified master regulators are in line with experimental evidence from the model bacterium E. coli. Together, our study demonstrates that simultaneous modelling of several target genes results in improved inference of GRNs and can be used as an alternative in different applications. Author summaryReconstruction of cellular networks based on snapshots of molecular profiles of the network components has been one of the key challenges in systems biology. In the context of reconstruction of gene regulatory networks (GRNs), this problem translates into inferring regulatory relationships between transcription factor coding genes and their targets based on, often small, number of expression profiles. While unsupervised nonlinear machine learning approaches have shown better performance than regularized linear regression approaches, the existing modeling strategies usually do predictions of regulators for one target gene at a time. Here, we ask if and to what extent the joint modeling of regulation for multiple targets leads to improvement of the accuracy of the inferred GRNs. To address this question, we proposed, implemented, and compared the performance of models cast as a blend between regularized multivariate regression and graphical models that combine the L2,1-norm with classical regularization techniques. Our results demonstrate that the proposed models, despite relying on linearity assumptions, show consistently good performance in comparison to existing, widely used alternatives.