Temperature Accelerated Molecular Dynamics with Soft-Ratcheting Criterion Orients Enhanced Sampling by Low-Resolution Information

Abstract

Many proteins exhibit an equilibrium between multiple conformations, some of them being characterized only by low-resolution information. Visiting all conformations is a demanding task for computational techniques performing enhanced but unfocused exploration of collective variable (CV) space. Otherwise, pulling a structure toward a target condition biases the exploration in a way difficult to assess. To address this problem, we introduce here the soft-ratcheting temperature-accelerated molecular dynamics (sr-TAMD), where the exploration of CV space by TAMD is coupled to a soft-ratcheting algorithm that filters the evolving CV values according to a predefined criterion. Any low resolution or even qualitative information can be used to orient the exploration. We validate this technique by exploring the conformational space of the inactive state of the catalytic domain of the adenyl cyclase AC from Bordetella pertussis. The domain AC gets activated by association with calmodulin (CaM), and the available crystal structure shows that in the complex the protein has an elongated shape. High-resolution data are not available for the inactive, CaM-free protein state, but hydrodynamic measurements have shown that the inactive AC displays a more globular conformation. Here, using as CVs several geometric centers, we use sr-TAMD to enhance CV space sampling while filtering for CV values that correspond to centers moving close to each other, and we thus rapidly visit regions of conformational space that correspond to globular structures. The set of conformations sampled using sr-TAMD provides the most extensive description of the inactive state of AC up to now, consistent with available experimental information.

Improved large-scale prediction of growth inhibition patterns using the NCI60 cancer cell line panel

“…The U-matrix value associated with a given neuron, , is defined as the average Euclidean distance between that neuron and its eight closest neighbours: , where is the set of neighbours and is the Euclidean distance between neurons. A distance threshold value was then applied to the U-matrix to define the contours of the compound clusters ( Cortes-Ciriano et al. , 2015c ).…”

Extended Phase-Space Methods for Enhanced Sampling in Molecular Simulations: A Review

“…In Yamamori and Kitao ( 2013 ), an interesting combination of TAMD and Replica Exchange Umbrella Sampling (REUS) was proposed. More recently, TAMD was combined with a soft-ratcheting algorithm to achieve a focused exploration of the CV space by relying on low-resolution or even qualitative experimental information (Cortes-Ciriano et al, 2015 ). Such an approach was successful in providing previously unavailable conformations of calmoduline-free adenyl cyclase toxin from the causative agent of whooping cough.…”

Exploring high‐dimensional free energy landscapes of chemical reactions

“…Maragliano et al and Lapelosa et al used this method to study the CO migration in myoglobin. Tzanov et al and Cortes‐Ciriano et al employed TAMD/d‐AFED to predict peptide conformations. A Monte‐Carlo version of the method was employed to study vacancy diffusion in crystals, wherein very complex CVs were used such as moments computed by the QM probability density of a pseudo‐probe particle.…”