Information theory, multivariate dependence, and genetic network inference

TL;DR

This paper introduces a novel information-theoretic approach using maximum entropy techniques and graphical notation to infer dependencies among multiple variables, especially useful in undersampled biological data scenarios.

Contribution

It presents a new method for inferring multivariate dependencies with maximum entropy and graphical notation, effective even with limited data.

Findings

Successfully tested on synthetic data

Uncovers dependencies in undersampled regimes

Potential applications in genetic and biological networks

Abstract

We define the concept of dependence among multiple variables using maximum entropy techniques and introduce a graphical notation to denote the dependencies. Direct inference of information theoretic quantities from data uncovers dependencies even in undersampled regimes when the joint probability distribution cannot be reliably estimated. The method is tested on synthetic data. We anticipate it to be useful for inference of genetic circuits and other biological signaling networks.