Reconstructability of evolutionary intermediates in generative epistatic landscapes

Roberto Netti, Martin Weigt · 2026

Read Reconstructability of evolutionary intermediates in generative epistatic landscapes by Roberto Netti, Martin Weigt (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-27983.txt.

Description

Evolutionary intermediates connect observed proteins. The step-by-step path that produced them is rarely recoverable, though. Roberto Netti and Martin Weigt ask in 2026 what the endpoints alone can reveal. Generative landscapes serve as controlled models of protein evolution. Reconstructions are checked against ground-truth simulated trajectories.

The best point prediction is not the most faithful reconstruction. Maximum-likelihood intermediates can be accurate residue by residue yet atypical in distribution. Conditional sampling captures the ensemble of plausible histories better. Predictability is limited by the landscape's shape. Regions of low mutability, being constrained, preserve information about the path. Open regions offer many routes and erase that memory. Sequence divergence alone also fails as a clock. Adding endpoint mutability places them more reliably. That limit is a property of the landscape, not of the method. The authors argue models should return realistic ensembles. They should also say when endpoints lack the information.

Catalog record

CREATORRoberto Netti, Martin Weigt
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE117 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.27983

Browse the full library · About the Tower of Babel