Metadynamic metainference: Enhanced sampling of the metainference ensemble using metadynamics.

Massimiliano Bonomi,Michele Vendruscolo,Carlo Camilloni

doi:10.1038/srep31232

Abstract

Accurate and precise structural ensembles of proteins and macromolecular complexes can be obtained with metainference, a recently proposed Bayesian inference method that integrates experimental information with prior knowledge and deals with all sources of errors in the data as well as with sample heterogeneity. The study of complex macromolecular systems, however, requires an extensive conformational sampling, which represents a separate challenge. To address such challenge and to exhaustively and efficiently generate structural ensembles we combine metainference with metadynamics and illustrate its application to the calculation of the free energy landscape of the alanine dipeptide.

Highlights

While metainference provides in principle a rigourous way to obtain ensembles of models consistent with the available experimental data, the actual generation of such models remains a computationally demanding task
We present metadynamic metainference (M&M), an approach that combines the ability of metainference to model heterogeneous systems by integrating noisy experimental data and prior knowledge of a system, with the enhanced sampling provided by PBMetaD
M&M is a modelling approach that combines metainference with metadynamics to model heterogeneous systems by integrating prior knowledge of the system with noisy experimental data. These two methods address very distinct problems, as metainference deals with the problem of accounting for statistical and systematic errors in experimental data collected on heterogeneous systems, while metadynamics deals with the problem of sampling the conformational space efficiently

Summary

Introduction

While metainference provides in principle a rigourous way to obtain ensembles of models consistent with the available experimental data, the actual generation of such models remains a computationally demanding task. The MetaD bias potential accellerates sampling by discouraging visiting regions of the CV space previously explored and provides an estimate of the free energy as a function of the selected CVs. The choice of a limited set of descriptors to capture all the slow modes of a system has always been proved to be a challenging task. We introduce the M&M theory and show its application to the prototypical case of alanine dipeptide in vacuo This system has become a well-established benchmark for many computational techniques, as it is characterized by multiple structural states that are significantly populated at room temperature and separated by high free-energy barriers

Methods

Results

Conclusion