Incorporating Nearest-Neighbor Site Dependence into Protein Evolution Models

Area of Science:

• Computational biology
• Molecular evolution
• Bioinformatics

Background:

• Protein evolutionary models are crucial for sequence alignment, homology, and phylogeny inference.
• Existing models often assume independent evolution between sites, which is unrealistic.
• Previous site-independent models for protein structural evolution improved alignments and phylogenetic inferences.

Purpose of the Study:

• To extend protein structural evolution models to account for correlated evolution among neighboring amino acid positions.
• To develop a spatiotemporal model for protein structure evolution.
• To reduce bias in evolutionary distance estimation and improve phylogenetic tree reconstruction.

Main Methods:

• Developed a site-dependent model (SDM) for protein structure evolution.
• The SDM uses a multivariate diffusion process convolved with a spatial birth-death process.
• Compared the SDM with site-independent models (SIM) for computational cost and analytical complexity.

Main Results:

• The site-dependent model (SDM) significantly reduces bias in estimated evolutionary distances.
• The SDM further improves phylogenetic tree reconstruction accuracy.
• Demonstrated bias in standard site-independent sequence evolution models using a site-dependent sequence evolution model.

Conclusions:

• Site-dependent models offer significant improvements over site-independent models for protein evolution.
• The developed spatiotemporal model provides a more realistic approach to understanding protein evolution.
• The SDM offers enhanced accuracy in phylogenetic analysis with minimal computational overhead.