Fluctuations and the limit of predictability in protein evolution

Area of Science:

• Evolutionary biology
• Statistical physics
• Computational biology

Background:

• Protein evolution is characterized by mutations across diverse timescales.
• Dynamical heterogeneity in evolution suggests correlations between mutations at different sites and times.
• Disordered systems in physics offer analogies for understanding these evolutionary dynamics.

Purpose of the Study:

• To quantify spatio-temporal correlations in protein evolution.
• To disentangle fluctuation sources: ancestral sequence vs. stochastic mutations.
• To investigate the influence of ancestral sequence properties on evolutionary trajectories and ancestral sequence reconstruction (ASR).

Main Methods:

• Simulated protein evolution using a data-driven energy landscape as a fitness proxy.
• Applied spatio-temporal correlation functions from disordered physical systems.
• Utilized a standard ancestral sequence reconstruction (ASR) algorithm.

Main Results:

• Fluctuations from the ancestral sequence dominate at shorter evolutionary timescales.
• A characteristic timescale exists for ancestral sequence information persistence.
• Stronger epistatic interactions in ancestors lead to longer persistence and impact evolutionary trajectories.
• Ancestral influence diminishes at longer timescales as sites evolve collectively.
• ASR algorithm performance is directly influenced by ancestral sequence properties, particularly epistasis.

Conclusions:

• Ancestral sequence properties, especially epistatic constraints, significantly shape early evolutionary dynamics.
• These properties impact the accuracy and performance of standard ancestral sequence reconstruction (ASR) methods.
• Understanding these correlations is key to deciphering evolutionary pathways and reconstructing ancestral states.