Extant Sequence Reconstruction: The Accuracy of Ancestral Sequence Reconstructions Evaluated by Extant Sequence...

Area of Science:

• Evolutionary biology
• Molecular evolution
• Biophysics

Background:

• Ancestral sequence reconstruction (ASR) infers ancient biomolecule properties and evolutionary mechanisms.
• The accuracy of ASR and the impact of evolutionary models remain largely unquantified due to the difficulty of verifying reconstructions against true ancestors.

Purpose of the Study:

• To evaluate the accuracy of ancestral sequence reconstruction (ASR) methods.
• To determine which evolutionary models are most suitable for ASR.
• To assess the reliability of common metrics for evaluating ASR quality.

Main Methods:

• Developed and applied extant sequence reconstruction (ESR), a cross-validation approach.
• Reconstructed each extant sequence within an alignment using ASR methodology.
• Compared ESR reconstructions to the original, known extant sequences to quantify accuracy.

Main Results:

• Average probability is a reliable accuracy measure only with accurate or overparameterized models.
• More accurate phylogenetic models can yield lower probability scores but higher biophysical similarity to true ancestors.
• Sampling from the reconstruction distribution can identify sequences with fewer errors than the single most probable reconstruction.

Conclusions:

• Model selection is critical for accurate ASR.
• ESR is a robust method for validating evolutionary models in ASR.
• Sampling sequence reconstructions aids in analyzing ancestral protein biophysical properties and validating resurrected proteins.