Scaling statistical multiple sequence alignment to large datasets.

Area of Science:

• Bioinformatics
• Computational Biology
• Evolutionary Biology

Background:

• Multiple sequence alignment is crucial in bioinformatics, with growing interest in large datasets.
• Accurate alignments often rely on stochastic models of sequence evolution.
• Existing methods struggle with large-scale datasets; BAli-Phy is limited to ~100 sequences.

Purpose of the Study:

• To extend the Bayesian phylogenetic analysis by inference (BAli-Phy) method for large-scale sequence alignments.
• To improve alignment and phylogenetic tree accuracy on datasets with thousands of sequences.
• To evaluate the performance of extended BAli-Phy against existing methods.

Main Methods:

• Incorporated BAli-Phy into PASTA and UPP, scaling strategies for alignment methods.
• Utilized simulated data with up to 10,000 sequences under various model conditions.
• Measured alignment and tree accuracy against ground truth from simulations.

Main Results:

• Extended BAli-Phy demonstrated superior performance on large datasets.
• Achieved significantly more accurate alignments and phylogenetic trees compared to current leading methods.
• The method showed robustness even with data divergent from statistical models.

Conclusions:

• Extensions of BAli-Phy using PASTA and UPP offer a significant advancement in large-scale sequence alignment.
• These enhanced methods produce more accurate alignments and phylogenetic trees.
• This approach has the potential to improve large-scale phylogenetic analyses beyond current capabilities.