ACES: A co-evolution simulator generates co-varying protein and nucleic acid sequences

Area of Science:

• Computational Biology
• Bioinformatics
• Molecular Evolution

Background:

• Predicting molecular interactions relies on identifying residue co-evolution in proteins and RNAs.
• Current benchmarking methods using real biological data are limited by unknown evolutionary constraints.
• Distinguishing direct from indirect co-evolutionary signals remains a challenge for existing algorithms.

Purpose of the Study:

• To develop a novel computational tool, aCES, for generating in silico simulated co-evolutionary sequences.
• To provide a reliable method for benchmarking co-evolution detection algorithms.
• To improve the understanding of residue interactions and their driving forces in molecular evolution.

Main Methods:

• Developed the aCES software tool to generate sequence alignments with specified conservation and co-evolutionary constraints.
• Utilized in silico generated alignments to benchmark multiple co-evolution detection algorithms.
• Analyzed the performance of algorithms in separating true co-evolutionary signals from background noise and indirect effects.

Main Results:

• aCES enables the creation of realistic, complex co-evolutionary datasets for method development.
• Benchmarking using aCES revealed the strengths and weaknesses of different co-evolution detection tools.
• Systematic tuning of constraints in aCES provided insights into the mechanisms driving residue co-evolution.

Conclusions:

• In silico sequence generation using aCES offers a superior alternative to real data for benchmarking co-evolution detection.
• Refined algorithms based on this approach can more accurately predict protein and RNA interactions.
• This work advances synthetic biology by enabling the engineering of novel macromolecular interactions.