A Mixture Model to Detect Edges in Sparse Co-expression Graphs

TL;DR

This paper introduces a novel mixture model for detecting edges in sparse gene co-expression networks, improving detection power and false discovery rate control without assuming specific network structures.

Contribution

It develops a three-component mixture model that effectively identifies gene network edges from co-expression data, outperforming existing methods.

Findings

The L2N mixture model outperforms other methods in edge detection power.

It effectively controls the false discovery rate.

The method makes no assumptions about the true network structure.

Abstract

In the early days of gene expression data, researchers have focused on gene-level analysis, and particularly on finding differentially expressed genes. This usually involved making a simplifying assumption that genes are independent, which made likelihood derivations feasible and allowed for relatively simple implementations. In recent years, the scope has expanded to include pathway and `gene set' analysis in an attempt to understand the relationships between genes. We develop a method to recover a gene network's structure from co-expression data, measured in terms of normalized Pearson's correlation coefficients between gene pairs. We treat these co-expression measurements as weights in the complete graph in which nodes correspond to genes. To decide which edges exist in the gene network, we fit a three-component mixture model such that the observed weights of `null edges' follow a normal distribution with mean 0, and the non-null edges follow a mixture of two lognormal distributions, one for positively- and one for negatively-correlated pairs. We show that this so-called L2N mixture model outperforms other methods in terms of power to detect edges, and it allows us to control the false discovery rate. Importantly, the method makes no assumptions about the true network structure.