RF-Net 2: Fast Inference of Virus Reassortment and Hybridization Networks

Abstract

Motivation: A phylogenetic network is a powerful model to represent entangled evolutionary histories with both divergent (speciation) and convergent (e.g., hybridization, reassortment, recombination) evolution. The standard approach to inference of hybridization networks is to (i) reconstruct rooted gene trees and (ii) leverage gene tree discordance for network inference. Recently, we introduced a method called RF-Net for accurate inference of virus reassortment and hybridization networks from input gene trees in the presence of errors commonly found in phylogenetic trees. While RF-Net demonstrated the ability to accurately infer networks with up to four reticulations from erroneous input gene trees, its application was limited by the number of reticulations it could handle in a reasonable amount of time. This limitation is particularly restrictive in the inference of the evolutionary history of segmented RNA viruses such as influenza A virus (IAV), where reassortment is one of the major mechanisms shaping the evolution of these pathogens. Results: Here we expand the functionality of RF-Net that makes it significantly more applicable in practice. Crucially, we introduce a fast extension to RF-Net, called Fast-RF-Net, that can handle large numbers of reticulations without sacrificing accuracy. Additionally, we develop automatic stopping criteria to select the appropriate number of reticulations heuristically and implement a feature for RF-Net to output error-corrected input gene trees. We then conduct a comprehensive study of the original method and its novel extensions and confirm their efficacy in practice using extensive simulation and empirical influenza A virus evolutionary analyses. Availability: RF-Net 2 is available at https://github.com/flu-crew/rf-net-2.

Inference of Phylogenetic Networks from Sequence Data using Composite Likelihood

“…Nevertheless, computational methods for inferring phylogenetic networks from genomic data are being actively developed under parsimony (Yu et al, 2013; Yan et al, 2022; Thomas et al, 2017), full (Yu et al, 2012, 2014) and composite likelihood (Yu and Nakhleh, 2015; Solís-Lemus and Ané, 2016, this study), Bayesian (Zhang et al, 2018; Flouri et al, 2020; Zhu et al, 2018; Wen et al, 2016), and other (Markin et al, 2022; Allman et al, 2019) frameworks. Among these, composite likelihood based methods are leading in overcoming scalability issues in network inference as they are shown to be more accurate than parsimony-based methods but more scalable than full likelihood methods (Solís-Lemus and Ané, 2016; Hejase and Liu, 2016).…”

“…Instead of deriving new theoretical results on properties of network space, we implemented the following five moves (one operation to traverse the tree space and four operations for network space) using the relevant functions available in the Julia package PhyloNetworks (Solís-Lemus et al, 2017). Similar operations are used in the heuristics implemented in PhyloNet (Yu et al, 2014;Yu and Nakhleh, 2015) and in the SubNet Prune and Regraft (Bordewich et al, 2017) edit operation implemented in RF-NET 2 (Markin et al, 2022). Note that the input and output networks for these moves are semi-directed networks, hence these networks lack directionality except for the reticulation edges.…”

“…Significant developments in phylogenetic network inference methods from multi-locus data have been made, but scalability remains a major challenge since the computational cost becomes large as the dataset size (i.e., number of individuals and/or sequence length) and/or topological complexity (i.e., number of reticulations) increase. One common approach to alleviate the computational burden is to summarize an input sequence alignment through estimation of gene trees for each locus prior to network estimation (e.g., PhyloNet (Than et al, 2008; Wen and Nakhleh, 2018), SNaQ (Solís-Lemus and Ané, 2016), NANUQ (Allman et al, 2019), RF-NET2 (Markin et al, 2022)). While data summarization improves efficiency, inference of networks from the gene trees is still computationally intensive (Hejase and Liu, 2016).…”