DeepAutoPIN: An automorphism orbits based deep neural network for characterizing the organizational diversity of protein interactomes across the tree of life

TL;DR

It is reported that orbit usage profiles (OUPs) of networks belonging to the three domains of life are contrastingly different not only at the domain level but also at the scale of phyla.

Abstract

The enormous diversity of life forms thriving in drastically different environmental milieus involves a complex interplay among constituent proteins interacting with each other. However, the organizational principles characterizing the evolution of protein interaction networks (PINs) across the tree of life are largely unknown. Here we study 4,738 PINs belonging to 16 phyla to discover phyla-specific architectural features and examine if there are some evolutionary constraints imposed on the networks' topologies. We utilized positional information of a network's nodes by normalizing the frequencies of automorphism orbits appearing in graphlets of sizes 2-5. We report that orbit usage profiles (OUPs) of networks belonging to the three domains of life are contrastingly different not only at the domain level but also at the scale of phyla. Integrating the information related to protein families, domains, subcellular location, gene ontology, and pathways, our results indicate that wiring patterns of PINs in different phyla are not randomly generated rather they are shaped by evolutionary constraints imposed on them. There exist subtle but substantial variations in the wiring patterns of PINs that enable OUPs to differentiate among different superfamilies. A deep neural network was trained on differentially expressed orbits resulting in a prediction accuracy of 85%.

Figures (4)

• Figure 1: Differentially expressed orbits (DEO): statistically significant orbits found to have Z-score$>2.58$ for at least $95 \%$ networks of the same phylum.
• Figure 2: Jaccard similarity (JC) matrices for function annotations, where each term represent an average JC value computed on pairwise JC obtained from $57$ orbit pairs of any two phyla: (a) JC matrix for protein family annotations. (b) JC matrix for protein domain annotations. (c) JC matrix for gene ontology annotations. (d) JC matrix for pathway annotations.
• Figure 3: Clustering of OUPs: (a) Average OUPs corresponding to each phyla clustered using distance correlation values. (b) All the $4,738$ OUPs clustered using neighbor-joining method, distance matrix computed using Bray-Curtis dissimilarity.
• Figure 4: Classification results of deep neural network (DeepAutoPIN) on $57$ dimensional OUPs: (a) Confusion matrix representing percentage of true and false instances predictions for every phyla. (b) ROC curves and Area under each ROC curve computed using one-vs-rest strategy for every phyla.