PIPE4: Fast PPI Predictor for Comprehensive Inter- and Cross-Species Interactomes

Abstract

The need for larger-scale and increasingly complex protein-protein interaction (PPI) prediction tasks demands that state-of-the-art predictors be highly efficient and adapted to inter-and cross-species predictions. Furthermore, the ability to generate comprehensive interactomes has enabled the appraisal of each PPI in the context of all predictions leading to further improvements in classification performance in the face of extreme class imbalance using the Reciprocal Perspective (RP) framework. We here describe the PIPE4 algorithm. Adaptation of the PIPE3/MP-PIPE sequence preprocessing step led to upwards of 50x speedup and the new Similarity Weighted Score appropriately normalizes for window frequency when applied to any inter-and cross-species prediction schemas. Comprehensive interactomes for three prediction schemas are generated: (1) cross-species predictions, where Arabidopsis thaliana is used as a proxy to predict the comprehensive Glycine max interactome, (2) interspecies predictions between Homo sapiens-HIV1, and (3) a combined schema involving both cross-and inter-species predictions, where both Arabidopsis thaliana and Caenorhabditis elegans are used as proxy species to predict the interactome between Glycine max (the soybean legume) and Heterodera glycines (the soybean cyst nematode). Comparing PIPE4 with the state-of-the-art resulted in improved performance, indicative that it should be the method of choice for complex PPI prediction schemas. The elucidation of protein-protein interaction (PPI) networks is central to molecular biology research. Necessary to producing mechanistic models of cellular processes, PPI networks additionally contribute to challenges such as the prediction of gene function 1-3 , identification of disease genes 4 , and pharmaceutical discovery 5,6. Computational PPI prediction techniques have been developed to supplement and guide wet-laboratory experimental work. The last decade has seen increased computational demand in both scale and complexity of PPI predictors. Predicting comprehensive interactomes (the set of all possible pairwise PPIs in or between proteomes) has only recently become possible with the advent of high-performance computing infrastructure and algorithmic optimizations. While methodologically diverse in their implementation, PPI prediction tools generally exploit information from the set of known PPIs (previously confirmed using classical wet-laboratory techniques) to determine whether any two query proteins will physically interact. The utility and scalability of any one method is subject to the information it leverages. Structure-based methods, at one extreme, require the three-dimensional (3D) characterization of each protein and therefore suffer from low coverage of the proteome. While useful to determining highly specific PPI networks, many methods require template-based modelling which tend to be computationally taxing 7-9. Furthermore, even with complete 3D structural information of each protein in an organism's proteome, the computational time comp...

Computational Prediction of the Comprehensive SARS-CoV-2 vs. Human Interactome to Guide the Design of Therapeutics

“…The fourth version of the Protein-protein Interaction Prediction Engine (PIPE4) was recently adapted to improve predictive performance for understudied organisms [6]. That is, species for which the proteome is known, however, the the number of experimentally validated PPIs involving the proteins of this organism is insufficient to train a model to generate the comprehensive interactome.…”

“…Due to the limited availability of known SARS-CoV-2 PPIs, we here use the PPIs from a collection of well-studied and evolutionarily similar proxy viruses to generate these cross-species predictions as depicted in Figure 1. The PIPE4 algorithm is particularly well-suited to cross-and inter-species PPI prediction schemas, given that the SW-scoring function appropriately normalizes the prevalence of sequence windows within each training and target species proteome [6].…”

“…Other labs have shared experimentally-determined and computationally-predicted structures of some of the viral proteins, and still others have shared epidemiological data. We hope to contribute to the scientific effort using the latest version of our sequence-based protein-protein interaction (PPI) predictor, PIPE4 [6] in combination with another state-of-the-art PPI predictor, denoted Scoring PRotein INTeractions (SPRINT) [7]. Finally, the PIPE-Sites algorithm was used to predict the sub-sequence regions with a high likelihood of mediating the physical interaction between two given pairs [8].…”

A multitask transfer learning framework for novel virus-human protein interactions

“…Virus protein sequences of different species share only little in common (Eid et al, 2016). Therefore, models trained for other human PPI (Li & Ilie, 2020; Sun et al, 2017; Li, 2020; Chen et al, 2019; Sarkar & Saha, 2019) or for other pathogen-human PPI (Sudhakar et al, 2020; Mei & Zhang, 2020; Dick et al, 2020; Li et al, 2014; Guven-Maiorov et al, 2019; Basit et al, 2018)(for which more data might be available) cannot be directly used for predictions for novel viral-human protein interactions.…”

A functional genomics approach in identifying the underlying gene for the E8 maturity locus in soybean (Glycine max)

“…E8 region located on chromosome 4, between physical positions of 7,000,000 bp -44,000,000 bp, containing approximately 1000 genes. The most updated version of soybean-PIPE (version 4)[68], was used to identify the top 200 interacting partners for each potential candidate gene. These top 200 interacting partners were analyzed using GO terms, identifying which, or how many, of these 200 interacting partners were involved in time of flowering and maturity.…”