WASCO: A Wasserstein-based Statistical Tool to Compare Conformational Ensembles of Intrinsically Disordered Proteins

Area of Science:

• Computational Biology
• Biophysics
• Structural Biology

Background:

• Intrinsically disordered proteins (IDPs) lack stable structures, necessitating ensemble models to capture their conformational diversity.
• Current methods require new statistical paradigms to analyze IDP conformational ensembles as probability distributions.

Purpose of the Study:

• To define a novel metric for comparing conformational ensembles of IDPs.
• To develop a method that accounts for the probabilistic nature and geometric properties of IDP conformational spaces.

Main Methods:

• Defined conformational ensembles as ordered sets of probability distributions.
• Developed residue-level metrics for local and global comparisons, integrating Euclidean and toroidal geometry.
• Incorporated data uncertainty for refined difference estimations.
• Defined an overall ensemble distance from residue-level differences.

Main Results:

• A new metric effectively detects differences between conformational ensembles at local and global scales.
• The method successfully compared ensembles from molecular dynamics (MD) simulations with varying force fields.
• The approach demonstrated utility in assessing MD simulation convergence and refinement against experimental data.

Conclusions:

• The developed statistical approach provides a robust framework for analyzing and comparing IDP conformational ensembles.
• This method enhances the understanding of protein dynamics and can be applied to refine structural models using experimental data.
• The Python-based tool facilitates applications in computational biology, including machine learning approaches.