[Paper Review] Global Alignment of Molecular Sequences via Ancestral State Reconstruction

Alexandr Andoni, Constantinos Daskalakis+2 more|arXiv (Cornell University)|Dec 14, 2009

DNA and Biological ComputingBiochemistry, Genetics and Molecular Biology3 citations

TL;DR

This paper introduces a rigorous model-based approach to global multiple sequence alignment using ancestral state reconstruction under stochastic indel models. It proposes a recursive algorithm that reconstructs the ancestral sequence from leaf sequences on a phylogenetic tree, achieving strong reconstruction guarantees under low mutation rates, marking the first such result with insertions and deletions in a tree-structured evolutionary model.

ABSTRACT

Molecular phylogenetic techniques do not generally account for such common evolutionary events as site insertions and deletions (known as indels). Instead tree building algorithms and ancestral state inference procedures typically rely on substitution-only models of sequence evolution. In practice these methods are extended beyond this simplified setting with the use of heuristics that produce global alignments of the input sequences--an important problem which has no rigorous model-based solution. In this paper we consider a new version of the multiple sequence alignment in the context of stochastic indel models. More precisely, we introduce the following {\em trace reconstruction problem on a tree} (TRPT): a binary sequence is broadcast through a tree channel where we allow substitutions, deletions, and insertions; we seek to reconstruct the original sequence from the sequences received at the leaves of the tree. We give a recursive procedure for this problem with strong reconstruction guarantees at low mutation rates, providing also an alignment of the sequences at the leaves of the tree. The TRPT problem without indels has been studied in previous work (Mossel 2004, Daskalakis et al. 2006) as a bootstrapping step towards obtaining optimal phylogenetic reconstruction methods. The present work sets up a framework for extending these works to evolutionary models with indels.

Motivation & Objective

• To address the lack of model-based solutions for global multiple sequence alignment in the presence of insertions and deletions (indels).
• To extend trace reconstruction from star-shaped phylogenies to general tree-structured evolutionary models with indels.
• To provide a theoretically sound framework for ancestral sequence reconstruction that accounts for both substitutions and indel events.
• To establish conditions under which ancestral sequences can be reconstructed with high probability using only leaf sequences and tree topology.

Proposed method

• Formulates a new trace reconstruction problem on trees (TRPT) that incorporates substitutions, deletions, and insertions via biased coin flips at each edge.
• Uses recursive reconstruction: for each internal node, aligns sequences from its children using anchor-based correlation tests to infer correct shifts and reconstruct the ancestral sequence.
• Employs anchor regions of length $ a = C\log n $ to compare subsequences across sequences and detect correct alignments via correlation thresholds.
• Applies adversarial reconstruction to bound error propagation, ensuring reconstructed sequences dominate true ancestral sequences under good events.
• Imposes conditions on mutation rates: $ p_i + p_d < O(1/k^{2/3} \log n) $ and $ (1-2p_s)^2 > O(d^{-1}\log d) $, to guarantee reconstruction accuracy.
• Uses probabilistic concentration bounds and event conditioning (e.g., $ \mathcal{E} $) to ensure high probability correctness across the tree.

Experimental results

Research questions

• RQ1 Can ancestral sequence reconstruction be rigorously achieved in the presence of insertions and deletions under a stochastic evolutionary model on a tree?
• RQ2 What conditions on mutation rates (substitution, deletion, insertion) enable accurate reconstruction of the root sequence from leaf sequences?
• RQ3 Can a recursive, model-based alignment procedure be designed that correctly identifies homologous sites across sequences evolving via indels?
• RQ4 How can correlation-based anchor matching be used to distinguish correct from incorrect sequence alignments in the presence of indels?

Key findings

• The algorithm achieves strong reconstruction of the ancestral sequence with high probability when $ p_i + p_d < O(1/k^{2/3} \log n) $ and $ (1-2p_s)^2 > O(d^{-1}\log d) $.
• Aligned anchor pairs exhibit correlation $ >(1-\delta)(1-2p_s)^2 $, while misaligned pairs have correlation $ < \delta $, enabling reliable alignment detection.
• The recursive reconstruction process dominates the adversarial reconstruction, ensuring correctness under good event conditions.
• The method provides the first rigorous trace reconstruction result on a tree with indels, extending prior work limited to substitution-only models.
• The algorithm ensures that the number of required events is polynomial in $ n $, making it efficient under the stated conditions.
• High-probability correctness is achieved by conditioning on a collection of good events $ \mathcal{E} $, including length stability and correlation consistency.