A Data-Driven Evolutionary Algorithm for Mapping Multibasin Protein Energy Landscapes.

Area of Science:

• Computational Biology
• Biophysics
• Structural Biology

Background:

• Proteins involved in proteinopathies are dynamic, switching structures to modulate function.
• Understanding protein structure-function relationships requires characterizing their conformation space.
• Current methods struggle to balance computational exploration and exploitation of energy surfaces.

Purpose of the Study:

• To develop a novel algorithm for comprehensive characterization of protein conformation space.
• To effectively balance computational resources for exploring and exploiting protein energy surfaces.
• To advance the modeling of multibasin proteins and understand sequence mutation impacts.

Main Methods:

• Utilizes evolutionary computation concepts to balance exploration and exploitation.
• Leverages deposited crystal structures to define a reduced, low-dimensional search space.
• Employs a multiscale technique to map samples to all-atom energy surfaces and novel strategies to avoid premature convergence.

Main Results:

• The algorithm effectively maps multibasin energy landscapes for various proteins.
• Demonstrates broad utility in advancing the modeling of multibasin proteins.
• Successfully applied to wildtype and variant protein sequences involved in proteinopathies.

Conclusions:

• The developed algorithm provides an effective balance for exploring protein conformation space.
• It serves as a crucial first step in understanding how sequence mutations affect protein misfunction.
• Offers the energy landscape as an intermediate link between protein sequence and function.