Finding functional associations between prokaryotic virus orthologous groups: a proof of concept

Nikolaos Pappas,Bas E Dutilh

doi:10.1186/s12859-021-04343-w

Abstract

BackgroundThe field of viromics has greatly benefited from recent developments in metagenomics, with significant efforts focusing on viral discovery. However, functional annotation of the increasing number of viral genomes is lagging behind. This is highlighted by the degree of annotation of the protein clusters in the prokaryotic Virus Orthologous Groups (pVOGs) database, with 83% of its current 9518 pVOGs having an unknown function.ResultsIn this study we describe a machine learning approach to explore potential functional associations between pVOGs. We measure seven genomic features and use them as input to a Random Forest classifier to predict protein–protein interactions between pairs of pVOGs. After systematic evaluation of the model’s performance on 10 different datasets, we obtained a predictor with a mean accuracy of 0.77 and Area Under Receiving Operation Characteristic (AUROC) score of 0.83. Its application to a set of 2,133,027 pVOG-pVOG interactions allowed us to predict 267,265 putative interactions with a reported probability greater than 0.65. At an expected false discovery rate of 0.27, we placed 95.6% of the previously unannotated pVOGs in a functional context, by predicting their interaction with a pVOG that is functionally annotated.ConclusionsWe believe that this proof-of-concept methodology, wrapped in a reproducible and automated workflow, can represent a significant step towards obtaining a more complete picture of bacteriophage biology.

Highlights

The field of viromics has greatly benefited from recent developments in metagenomics, with significant efforts focusing on viral discovery
It is becoming clear that bacteriophage genomes may encode functions that were previously thought to be carried out exclusively by cellular organisms, such as auxiliary metabolic genes involved in photosynthesis and carbon metabolism [6] or sulfur and nitrogen cycling [7]
Interaction datasets A discretely labeled ground truth dataset of interacting (1) and potentially non-interacting (0) protein pairs for supervised machine leaning with Random Forest [22] was constructed as follows: profile Hidden Markov Models (HMMs) of bacteriophage protein families and their functional annotations were retrieved from the prokaryotic virus orthologous group (pVOG) database [12]

Summary

Introduction

The field of viromics has greatly benefited from recent developments in metagenomics, with significant efforts focusing on viral discovery. Functional annotation of the increasing number of viral genomes is lagging behind This is highlighted by the degree of annotation of the protein clusters in the prokaryotic Virus Orthologous Groups (pVOGs) database, with 83% of its current 9518 pVOGs having an unknown function. The vast diversity across all environments of viruses that infect bacteria and archaea, together referred to as bacteriophages, has long been postulated [1]. New lineages are being discovered in different environments, such as crAssphage [2] and megaphages [3] in the Pappas and Dutilh BMC Bioinformatics (2021) 22:438 human gut or novel Vibrionaceae-infecting phages with relatively wide host-range in marine biomes [4], shedding light into the unexplored component of the virosphere’s diversity [5]. It is becoming clear that bacteriophage genomes may encode functions that were previously thought to be carried out exclusively by cellular organisms, such as auxiliary metabolic genes involved in photosynthesis and carbon metabolism [6] or sulfur and nitrogen cycling [7]

Methods

Results

Discussion

Conclusion