TL;DR
This paper introduces virDTL, a novel computational workflow for analyzing viral evolution, specifically addressing the challenge of frequent recombination in viruses like SARS-CoV-2. By adapting methods from prokaryotic horizontal gene transfer analysis and accounting for multiple sources of uncertainty, virDTL can identify ancestral recombination events, offering a more comprehensive evolutionary history for rapidly evolving viruses.
What Problem Does This Paper Address?
Understanding the evolutionary history of rapidly evolving viruses, such as SARS-CoV-2, is critical for tracking and preventing pandemics. However, viruses frequently undergo recombination, a process where genetic material is exchanged between different viral genomes. This recombination complicates traditional phylogenetic analysis, making it difficult to accurately reconstruct evolutionary relationships and identify ancestral events. Existing methods often struggle with inferring these ancestral recombinations, especially when multiple recombination events occur within a lineage or when dealing with the inherent uncertainties in phylogenetic inference. This paper aims to bridge this gap by providing a robust computational framework to tackle these challenges.
Methods
The core of this work is the virDTL (viral Duplication-Transfer-Loss) workflow. It's built upon Duplication-Transfer-Loss (DTL) reconciliation, a technique previously used for inferring horizontal gene transfer in prokaryotes. Key innovations and steps in virDTL include:
- Strain Tree Reconstruction and Selection: Identifying non-recombinant or minimally recombinant genomic regions to build a reliable "strain tree" that represents the evolutionary history of the viral strains.
- Gene Tree Reconstruction and Error-Correction: Reconstructing gene trees for individual gene families and then correcting them for phylogenetic reconstruction errors and uncertainty. This step uses RAxML for initial tree building and TreeFix-DTL for error correction.
- Gene Tree Rooting: Addressing the uncertainty in rooting gene trees by employing multiple rooting approaches (OptRoot and Minimum Ancestor Deviation) to assess their impact on the final inferences.
- Phylogenetic Reconciliation Analysis: Using RANGER-DTL to perform DTL reconciliation, accounting for uncertainty by sampling multiple optimal reconciliations for each gene tree and aggregating inferences.
- Strain Tree Dating and HGT Evaluation: Dating the strain tree to evaluate the temporal consistency of inferred Horizontal Gene Transfers (HGTs).
virDTL is designed to explicitly account for various sources of inference uncertainty, including strain tree construction, gene tree estimation and rooting, and the reconciliation process itself.
Datasets
The study utilized 54 Sarbecovirus genomes from 4 host species: 42 from bats, 5 from pangolins, 5 from civet cats, and 2 from humans. These genomes were obtained from the NCBI sequence database. The authors also reference and compare their findings with results from previous studies that used different datasets of coronaviruses.
Key Insights and Ideas
- Recombination is Pervasive in Sarbecoviruses: The analysis strongly suggests that recombination is a significant driver of evolution within the Sarbecovirus subgenus, impacting the evolutionary history of SARS-CoV-2.
- Ancestral Recombinations are Crucial: Identifying ancestral recombination events, not just recent ones between sampled strains, is key to understanding the full evolutionary picture.
virDTL's ability to infer these ancestral events is a major advancement. - Pangolin-SARS-CoV-2 Link is Recombinatorial: The observed genetic similarity between SARS-CoV-2 and pangolin coronaviruses is likely due to a recombination event between the ancestors of SARS-CoV-2/RaTG13 and pangolin strains, rather than pangolins being a direct intermediate host.
- Spike and Nucleocapsid Genes are Hotspots: The Spike (S) and Nucleocapsid (N) genes were identified as being frequently involved in recombination events, highlighting their importance in viral adaptation and evolution.
- Methodological Rigor Matters: The paper emphasizes how accounting for multiple sources of uncertainty (gene tree error, rooting, reconciliation optima) leads to more robust and reliable inference of HGTs.
Key Results
- Extensive HGTs Identified: The study identified 588 supported HGT events across all gene families, with 226 being leaf-to-leaf (between sampled strains) and 362 involving ancestral nodes.
- SARS-CoV-2 Lineage Insights: Eight highly supported HGTs involving the SARS-CoV-2 lineage were identified, including key events in the Spike and Nucleocapsid genes.
- Validation with SimPlot: A subset of inferred leaf-to-leaf HGTs was validated using SimPlot, demonstrating consistency between the reconciliation-based approach and sequence similarity analysis.
- Recombination Across Gene Boundaries: Statistical tests suggest that HGTs in adjacent gene families are likely part of a single, larger recombination event, indicating the potential to identify multi-gene recombination regions.
- Paucity of HGTs in SARS-CoV-2 Clade: Interestingly, the ancestors and close relatives of SARS-CoV-2 showed a lower frequency of HGTs compared to the rest of the Sarbecovirus subgenus, which the authors suggest might be due to sampling effects.
Limitations and Future Work
- Gene-by-Gene Analysis:
virDTLanalyzes each gene family separately, meaning it cannot directly infer recombination events that affect only parts of genes. Future work could address this by incorporating window-based analyses or by combining inferred HGTs with gene adjacency information to identify larger recombination regions. - Ignoring Low-Support HGTs: The current analysis focuses on highly supported HGTs, potentially overlooking important evolutionary insights from low-support events, especially in short genes which often have less phylogenetic signal. A closer analysis of these low-support HGTs is suggested.
- Directional Uncertainty: While
virDTLaccounts for uncertainty, inferring the precise direction of some HGTs can remain challenging, particularly in ancestral cases or when species tree resolution is limited. - Future Directions: The authors suggest that their approach can be extended to investigate other rapidly evolving viruses and to better understand the mechanisms and consequences of recombination in viral evolution. The
virDTLprotocol and scripts are made freely available for further research.
Summary updated