WASCO

WASCO compares conformational ensembles of intrinsically disordered proteins (IDPs) using a Wasserstein-based statistical framework to quantify residue-level and global differences between ensembles.

Key Features:

• Statistical Framework: Treats conformational ensembles as ordered sets of probability distributions to enable rigorous local and global statistical comparisons.
• Wasserstein-based Metric: Employs a Wasserstein-based metric to detect residue-level differences while integrating conformational-space geometry and supporting comparisons on three-dimensional Euclidean spaces and two-dimensional flat tori.
• Uncertainty Integration: Accounts for inherent data uncertainty to provide refined estimations of ensemble differences and aggregates residue-level distances into an overall ensemble distance.

Scientific Applications:

• Molecular Dynamics Simulations: Compare conformational ensembles produced using different force fields or before and after refinement with experimental data.
• Convergence Assessment: Assess convergence of molecular dynamics simulations by quantifying whether sampled ensembles differ across time or between replicas.
• Machine Learning Integration: Provide robust statistical comparisons of protein ensembles to support or evaluate machine-learning-based models.

Methodology:

Treats each ensemble as an ordered set of probability distributions and applies a Wasserstein-based metric on three-dimensional Euclidean spaces and two-dimensional flat tori to compute residue-level distances, incorporates uncertainty in estimations, and aggregates residue-level differences into an overall distance between ensembles.