[Paper Review] Recovering the tree-like trend of evolution despite extensive lateral genetic transfer: A probabilistic analysis

Sébastien Roch, Sagi Snir|arXiv (Cornell University)|Jun 15, 2012

Genomics and Phylogenetic Studies39 references4 citations

TL;DR

This paper presents a probabilistic analysis showing that the species tree can be accurately reconstructed even with a nearly linear number of lateral gene transfer (LGT) events per gene tree, using a robust, computationally efficient method based on quartet aggregation and median distance estimation. The key result is that the tree-like evolutionary signal remains recoverable under high LGT rates, provided gene tree topologies and distances are known or estimable with sufficient sequence length.

ABSTRACT

Lateral gene transfer (LGT) is a common mechanism of non-vertical evolution where genetic material is transferred between two more or less distantly related organisms. It is particularly common in bacteria where it contributes to adaptive evolution with important medical implications. In evolutionary studies, LGT has been shown to create widespread discordance between gene trees as genomes become mosaics of gene histories. In particular, the Tree of Life has been questioned as an appropriate representation of bacterial evolutionary history. Nevertheless a common hypothesis is that prokaryotic evolution is primarily tree-like, but that the underlying trend is obscured by LGT. Extensive empirical work has sought to extract a common tree-like signal from conflicting gene trees. Here we give a probabilistic perspective on the problem of recovering the tree-like trend despite LGT. Under a model of randomly distributed LGT, we show that the species phylogeny can be reconstructed even in the presence of surprisingly many (almost linear number of) LGT events per gene tree. Our results, which are optimal up to logarithmic factors, are based on the analysis of a robust, computationally efficient reconstruction method and provides insight into the design of such methods. Finally we show that our results have implications for the discovery of highways of gene sharing.

Motivation & Objective

• To determine the maximum tolerable rate of lateral gene transfer (LGT) before the tree-like evolutionary signal is irreversibly obscured.
• To evaluate the effectiveness of distance-based and quartet-based phylogenetic reconstruction methods in the presence of extensive LGT.
• To provide a theoretical foundation for designing robust, computationally efficient methods that extract the underlying species tree from conflicting gene trees.
• To investigate the impact of gene tree branch length assumptions and sequence length on reconstruction accuracy.
• To explore implications for detecting 'highways of gene sharing'—preferential LGT between specific clades.

Proposed method

• Uses a stochastic model of random LGT events distributed across the species phylogeny, with rates bounded by a parameter Λ.
• Employs a median-based reconstruction algorithm (MT) that aggregates quartet topologies and evolutionary distances across genes to infer the species tree.
• Applies concentration bounds and probabilistic analysis to show that, with high probability, the median of gene tree distances reflects the true species distances despite LGT.
• Analyzes the weight of paths in the species tree under the Yule and bounded-rates models to bound the effect of LGT on distance estimation.
• Considers sequence length requirements by modeling sequence evolution under the GTR model and using log-det distances to estimate evolutionary distances.
• Uses theoretical tools from probability and stochastic processes, including extreme value theory for minimum branch lengths and concentration inequalities for distance estimation.

Experimental results

Research questions

• RQ1 What is the maximum number of lateral gene transfer events per gene tree that still allows accurate reconstruction of the species phylogeny?
• RQ2 Can a simple, computationally efficient method recover the species tree when LGT events are randomly distributed across the species tree?
• RQ3 How does the number of genes and sequence length affect the accuracy of species tree reconstruction under high LGT rates?
• RQ4 To what extent do 'highways of gene sharing'—preferential LGT between specific clades—affect the recoverability of the species tree?
• RQ5 Is the tree-like signal in prokaryotic evolution robust to LGT, even when the number of transfers is nearly linear in the number of branches?

Key findings

• The species phylogeny can be reconstructed with high probability even when the number of LGT events per gene tree is as high as Ω(n), under the Yule and bounded-rates species tree models.
• The median of gene tree evolutionary distances is concentrated around the true species distances, provided the number of genes is polynomial in n and sequence lengths are polynomial in n.
• The method achieves reconstruction under a nearly linear number of LGT events, which exceeds known empirical estimates of LGT frequency in prokaryotes.
• The theoretical bound on tolerable LGT is tight up to logarithmic factors, indicating that the method is nearly optimal.
• The shortest branch length in the species tree is typically 1/poly(n), and sequence lengths scaling polynomially in n are sufficient for accurate distance estimation under the GTR model.
• Constraining LGT to closely related species significantly improves the feasibility of tree reconstruction, suggesting that local transfer events are less disruptive than random global transfers.