Alignment-Integrated Reconstruction of Ancestral Sequences Improves Accuracy

Area of Science:

• Computational Biology
• Molecular Evolution
• Bioinformatics

Background:

• Ancestral sequence reconstruction (ASR) infers ancient protein sequences using extant sequences, phylogenies, and evolutionary models.
• ASR allows direct study of sequence evolution's impact on protein structure and function.
• Previous research indicates phylogenetic uncertainty has minimal impact, while alignment uncertainty strongly affects ASR accuracy.

Purpose of the Study:

• To demonstrate how sequence alignment errors impact ASR accuracy across various evolutionary scenarios.
• To introduce and evaluate an alignment-integrated ASR approach to mitigate alignment uncertainty.
• To assess the effect of improved ASR on structural and functional inferences of ancestral proteins.

Main Methods:

• Evaluated ASR accuracy under realistic and simplified evolutionary scenarios with alignment errors.
• Developed and implemented an alignment-integrated ASR method combining multiple sequence alignments.
• Compared the performance of the new method against traditional ASR and structure-guided alignment.

Main Results:

• Sequence alignment errors were shown to produce significant errors in ASR, affecting structural and functional property estimations.
• The alignment-integrated ASR approach substantially improved ASR accuracy and downstream inferences.
• The novel method's performance was comparable to highly accurate structure-guided alignment approaches.

Conclusions:

• Errors in protein sequence alignment can undermine the reliability of ancestral sequence reconstruction studies.
• Integrating alignment uncertainty into ASR methods is crucial for improving accuracy and downstream analyses.
• Future ASR studies should adopt strategies to address alignment uncertainty for more reliable results.