Prokaryotic virus host predictor: a Gaussian model for host prediction of prokaryotic viruses in metagenomics

Abstract

Background Viruses are ubiquitous biological entities, estimated to be the largest reservoirs of unexplored genetic diversity on Earth. Full functional characterization and annotation of newly discovered viruses requires tools to enable taxonomic assignment, the range of hosts, and biological properties of the virus. Here we focus on prokaryotic viruses, which include phages and archaeal viruses, and for which identifying the viral host is an essential step in characterizing the virus, as the virus relies on the host for survival. Currently, the method for determining the viral host is either to culture the virus, which is low-throughput, time-consuming, and expensive, or to computationally predict the viral hosts, which needs improvements at both accuracy and usability. Here we develop a Gaussian model to predict hosts for prokaryotic viruses with better performances than previous computational methods. Results We present here Prokaryotic virus Host Predictor (PHP), a software tool using a Gaussian model, to predict hosts for prokaryotic viruses using the differences of k-mer frequencies between viral and host genomic sequences as features. PHP gave a host prediction accuracy of 34% (genus level) on the VirHostMatcher benchmark dataset and a host prediction accuracy of 35% (genus level) on a new dataset containing 671 viruses and 60,105 prokaryotic genomes. The prediction accuracy exceeded that of two alignment-free methods (VirHostMatcher and WIsH, 28–34%, genus level). PHP also outperformed these two alignment-free methods much (24–38% vs 18–20%, genus level) when predicting hosts for prokaryotic viruses which cannot be predicted by the BLAST-based or the CRISPR-spacer-based methods alone. Requiring a minimal score for making predictions (thresholding) and taking the consensus of the top 30 predictions further improved the host prediction accuracy of PHP. Conclusions The Prokaryotic virus Host Predictor software tool provides an intuitive and user-friendly API for the Gaussian model described herein. This work will facilitate the rapid identification of hosts for newly identified prokaryotic viruses in metagenomic studies.

Global overview and major challenges of host prediction methods for uncultivated phages

“…While HostPhinder predicts phage-host interactions by comparing 16-mers composition between query phage and known phage genome sequences [39], most tools instead compare k-mer frequencies between the query phage and a host reference genome database, based on the assumption that virus and host genomes often display similar sequence composition bias and k-mer frequencies patterns (Table 1). Prokaryotic virus host Predictor (PHP) [40] and VirHostMatcher (VHM) [41] respectively compare 4-mer and 6-mer frequency vectors between the query phage and a database of reference host genomes. "Who Is the Host?"…”

“…In contrast, the most recent tools leverage advances in machine learning to identify reliable predictions, including Gaussian models [40], neighborhood regularized logistic matrix factorization [43], and deep convolution neural network [45]. A critical and challenging aspect of these techniques is the establishment of robust and balanced training and test sets which should ideally represent a diverse range of viruses, hosts, and virus-host interactions, to avoid over-estimating the performance of these tools, i.e.…”

“…Overall, alignment-free tools tend to offer a higher recall but lower precision than alignment-based tools (Table 1). Across the different tools, accuracy values of 30%-70% have been reported [17,[40][41][42][43]45] but there is currently no systematic benchmark of these different tools on a common set of genomes as was done in ref. [17] for alignment-based approaches, so that a direct comparison of these tools' performance is challenging to establish.…”

Computational Viromics: Applications of the Computational Biology in Viromics Studies

“…Host predictions of eukaryotic viruses have been usually conducted based on viral sequences alone, such as those for influenza viruses and coronaviruses (Xu et al 2017 ; Tian, 2020 ); while those of prokaryotic viruses have been usually conducted based on the similarity of sequence features or sequences between viruses and hosts. At present, two kinds of computational methods have been developed to predict prokaryotic virus hosts based on genomic sequences (Edwards et al 2016 ; Ahlgren et al 2017 ; Galiez et al 2017 ; Lu et al 2021 ). The first kind of methods rely on the sequence similarity search between the query viruses and the candidate host genomes since viruses and their hosts may share the same genes and/or short nucleotide sequences such as the spacer sequences used in CRISPR systems (Edwards et al 2016 ).…”

“…However, they can be only used for a small proportion of viruses since only some viruses have sequence similarities with their hosts (Edwards et al 2016 ). Another kind of methods can predict the viral hosts based on the sequence composition similarity between viruses and their hosts, such as the Prokaryotic virus Host Predictor (PHP) (Lu et al 2021 ), VirHostMatcher (Ahlgren et al 2017 ) and WIsH (Galiez et al 2017 ). Although the latter kind of method predicts viral hosts with lower accuracy than the former, they can be used for any prokaryotic viruses.…”