PASTA with many application-aware optimization criteria for alignment based phylogeny inference.

Area of Science:

• Bioinformatics
• Computational Biology
• Phylogenetics

Background:

• Multiple sequence alignment (MSA) is crucial for bioinformatics analyses like phylogeny estimation.
• PASTA is a scalable and accurate MSA method that iteratively co-estimates MSAs and phylogenetic trees using a maximum likelihood (ML) score.
• The ML score, while effective, isn't a definitive criterion for accurate phylogenetic tree reconstruction.

Purpose of the Study:

• To improve MSA and phylogenetic tree accuracy by integrating multiple application-aware objectives into the PASTA framework.
• To develop a multi-objective (MO) framework, PMAO, that leverages PASTA to produce a set of high-quality, equivalent solutions.
• To provide tools for selecting optimal trees from the generated solution set.

Main Methods:

• Developed a multi-objective (MO) framework named PMAO.
• Integrated four application-aware objectives alongside the ML score within PASTA.
• Implemented a component to generate a smaller, high-quality subset of solutions from the PMAO output.
• Explored summarizing solutions to obtain a single high-quality tree.

Main Results:

• Experimental analysis on a biological benchmark showed that PMAO generates significantly better phylogenetic trees than standalone PASTA.
• The PMAO framework produces a diverse set of high-quality solutions, offering improved tree reconstruction.
• The added component effectively narrows down the solution set for easier selection by domain experts.

Conclusions:

• PMAO enhances phylogenetic tree accuracy by incorporating multiple objectives into the PASTA algorithm.
• The multi-objective approach yields a superior tree-space compared to traditional methods.
• PMAO offers practical tools for selecting optimal phylogenetic trees, potentially leading to more accurate downstream bioinformatics analyses.