Augmented Sparse Reconstruction of Protein Signaling Networks

TL;DR

This paper introduces augmented sparse reconstruction, a novel mathematical method for accurately identifying protein signaling networks from limited and noisy trajectory data, demonstrated on the EGFR network.

Contribution

The paper presents a new augmented sparse reconstruction technique that improves network inference accuracy in noisy, limited data scenarios, specifically for protein signaling pathways.

Findings

High accuracy in detecting network links despite noise

Effective in small data sets with limited trajectories

Potential for future therapeutic applications

Abstract

The problem of reconstructing and identifying intracellular protein signaling and biochemical networks is of critical importance in biology today. We sought to develop a mathematical approach to this problem using, as a test case, one of the most well-studied and clinically important signaling networks in biology today, the epidermal growth factor receptor (EGFR) driven signaling cascade. More specifically, we suggest a method, augmented sparse reconstruction, for the identification of links among nodes of ordinary differential equation (ODE) networks from a small set of trajectories with different initial conditions. Our method builds a system of representation by using a collection of integrals of all given trajectories and by attenuating block of terms in the representation itself. The system of representation is then augmented with random vectors, and minimization of the 1-norm is used to find sparse representations for the dynamical interactions of each node. Augmentation by random vectors is crucial, since sparsity alone is not able to handle the large error-in-variables in the representation. Augmented sparse reconstruction allows to consider potentially very large spaces of models and it is able to detect with high accuracy the few relevant links among nodes, even when moderate noise is added to the measured trajectories. After showing the performance of our method on a model of the EGFR protein network, we sketch briefly the potential future therapeutic applications of this approach.