Sequence variability of proteins evolutionarily constrained by solution-thermodynamic function.

Focusing on silk fibroin and hemoglobin molecules as templates, we model protein homolog dispersal across sequence-fitness landscapes determined by solution thermodynamics. Landscapes are constructed by inspecting an idealized theoretical phase topology associated with sequence length and hydrophobic-polar composition, comprising liquid-liquid phase separation, gelation and liquid crystalline self-assembly. We then calculate the distribution of homologs in sequence space as steady states of a simple mutation-selection dynamics. The results are consistent with Swiss-Prot bioinformatic data.