Biomolecular simulation for the end-stage refinement of nucleic acid structure

点击翻译按钮获取中文摘要

Project Summary / Abstract Despite significant advances in the performance and reliability of biomolecular simulation approaches, non- helical nucleic acid structures are proving difficult to fully and accurately model with currently available biomolecular force fields. This research aims to assess, validate, and improve force fields for nucleic acids, and also to better understand the true conformational ensemble of model DNA and RNA systems by best fits to NMR experiment using Maximum Entropy methods. This includes characterization of not only the dominant conformations, but also excited states or low population states. Model systems include RNA dinucleotides, RNA tetranucleotides, RNA tetraloops, DNA mini-dumbbells, and NMR-derived NMR structures that are known to populate multiple structures or excited states. Methods employed include state-of-the-art multi-dimensional replica-exchange molecular dynamics (M-REMD) simulations and various NMR approaches to provide more insight into new tetranucleotides both in structure (NOE, J-coupling, etc.) and for excited states via NMR relaxation and NMR relaxation dispersion experiments. The M-REMD code will be extended to allow asynchronous ensemble instances (for greater efficiency and queue backfill) and adaptivity and steering with on-the-fly analyses to assess convergence and where more/fewer ensemble instances are needed. Various approaches to force field improvement include surrogate methods for re-weighting converged MD trajectory ensembles and project free energy surfaces with Multistate Bennett Acceptance Ratio methods as a function of force field parameter change, where possible, via parameter scanning, via fits to high level base-base interaction energies from high level QM, and small model compound fits to liquid densities and other methods. In addition, the group will work with the Open Force Field Initiative to adapt, assess, and validate their small molecule force fields to nucleic acids. Using M-REMD methods the team has proven the ability to fully sample the conformational ensemble of the proposed model systems with large-scale computation on GPUs with multiple different force fields. The group has considerable experience in large-scale simulation of DNA and RNA and a proven track record of collaboration and dissemination of research findings and results. Development of better methods for simulation of the structure, dynamics and interactions of nucleic acids provides the means to approach drug- ability and the design of novel therapeutics, and also to provide greater insight into the role of structure, dynamics and conformational change in function which at a basic science level provides unique capabilities that could have considerable health relevance if the methods are made to function correctly and accurately.