Erroneous Rates and False Statistical Confirmations from Infrequent Metadynamics and Other Equivalent Violations of...

Area of Science:

• Computational Chemistry
• Molecular Dynamics Simulations
• Biophysics

Background:

• Hyperdynamics and adaptive biasing potentials are used to accelerate molecular simulations.
• These methods often assume bias potentials are zero on free-energy barriers of interest.
• This assumption is critical for the accuracy of simulation results.

Purpose of the Study:

• To demonstrate the flaw in the implicit assumption of hyperdynamics and biasing potentials.
• To investigate the conditions under which hyperdynamics in collective variables may fail.
• To evaluate the reliability of statistical tests used to validate hyperdynamics simulations.

Main Methods:

• Theoretical analysis of hyperdynamics and biasing potential assumptions.
• Application of the one-sample Kolmogorov-Smirnov test.
• Demonstration using the alanine dipeptide benchmark system.
• Analysis of a protein conformational change simulation.

Main Results:

• The implicit assumption that bias potentials are zero on prominent barriers is flawed.
• Hyperdynamics in collective variables can fail if hidden barriers exist in the initial state.
• The one-sample Kolmogorov-Smirnov test is inadequate for detecting these hyperdynamics failures.
• Nonzero bias potentials on hidden barriers contradict the expected Poisson distribution of escape times.

Conclusions:

• The current standards for trustworthy hyperdynamics simulations in collective variables are insufficient.
• Hidden free-energy barriers pose a significant challenge to the accuracy of these methods.
• Further development is needed to ensure the reliability of accelerated molecular dynamics techniques.