Robust Multi-view Co-expression Network Inference

TL;DR

This paper introduces a robust method for inferring gene co-expression networks from multiple transcriptome studies, effectively handling noise, batch effects, and spurious correlations to improve network accuracy.

Contribution

The authors propose a novel high-dimensional graph inference approach based on multivariate t-distributions and an EM algorithm, enhancing robustness over existing methods.

Findings

Improved accuracy in recovering gene co-expression networks.

Effective handling of batch effects and noise in multi-study data.

Demonstrated superiority over baseline methods on synthetic and real data.

Abstract

Unraveling the co-expression of genes across studies enhances the understanding of cellular processes. Inferring gene co-expression networks from transcriptome data presents many challenges, including spurious gene correlations, sample correlations, and batch effects. To address these complexities, we introduce a robust method for high-dimensional graph inference from multiple independent studies. We base our approach on the premise that each dataset is essentially a noisy linear mixture of gene loadings that follow a multivariate $t$-distribution with a sparse precision matrix, which is shared across studies. This allows us to show that we can identify the co-expression matrix up to a scaling factor among other model parameters. Our method employs an Expectation-Maximization procedure for parameter estimation. Empirical evaluation on synthetic and gene expression data demonstrates our method's improved ability to learn the underlying graph structure compared to baseline methods.