Covariance of maximum likelihood evolutionary distances between sequences aligned pairwise.

Area of Science:

• Molecular Evolution
• Bioinformatics
• Computational Biology

Background:

• Estimating distances between biological sequences is vital in molecular evolution.
• Maximum Likelihood (ML) is commonly used, but covariance estimators for independently aligned pairs are lacking.
• Current methods are computationally expensive for large-scale analyses requiring repeated Multiple Sequence Alignments (MSAs).

Purpose of the Study:

• To introduce a novel estimator for the covariance of distances derived from pairwise sequence alignments.
• To address the need for efficient covariance estimation in large-scale analyses where MSAs are prohibitive.
• To enhance processes like phylogenetic tree building and orthology inference.

Main Methods:

• Developed a new estimator for the covariance of pairwise sequence distances.
• Conducted extensive Monte Carlo simulations to analyze the estimator's performance.
• Compared the new estimator against the established ML distance variance estimator.

Main Results:

• The introduced covariance estimator demonstrates performance comparable to the ML variance estimator.
• The estimator shows no significant bias for sequence divergences below 150 PAM units (~29% sequence identity).
• Covariances may be underestimated at higher divergence levels, mirroring ML variance underestimation.

Conclusions:

• The new covariance estimator is a valuable tool for molecular evolution studies.
• It integrates effectively with ML variance estimators to construct covariance matrices.
• The estimator provides reliable results for a significant range of sequence divergences.