Recently, a new strategy for estimating the binding free energy of an SBP to a target surface via computational screening was proposed with preliminary results suggesting a 90% reduction in computational expense over traditional methods [1] . The new method involves a thorough analysis of individual amino acid binding free energies to a surface and then candidate SBPs can be screened through simple unbiased simulations of surface-bound peptides run for modest durations. The expedited acquisition of a binding free energy, the target observable, is achieved through clever approximations based on statistical mechanics principles that emphasize statistical sampling of bound states instead of explicitly monitoring binding/unbinding events. This work seeks to demonstrate the efficacy of this method for predicting the preferences of SBPs for different crystal facets of gold (and in the future, calcium carbonate), which will be validated against an experimentally measured database.
Preliminary results demonstrate agreement with the established paradigm, wherein peptides with a strong predicted binding affinity tend to nucleate nanoparticles experimentally. Interestingly, however, we observe that this is not uniformly the case, and peptides that bind to gold surfaces exceptionally strongly (>20 kT) in silico do not nucleate well experimentally. We hypothesize that this is a competitive effect wherein SBPs interfere with nucleation by preferentially occupying facets for growth. We intend to apply the described simulation workflow to duplicate these findings for approximately 100 peptides in our in-house experimental database for gold and use computational measurements on per-residue binding statistics to generate candidate peptides with optimized properties.
[1] Qi, X., & Pfaendtner, J. (2024). High-Throughput Computational Screening of Solid-Binding Peptides. J. Chem. Theory Comput., 20(7), 2959-2968.
Statement of Research Interests:
My research expertise lies in the development and application of theoretical and computer simulation techniques toward studying fluid and materials properties. Of particular interest are molecular-based models that explicitly represent the interactions between microscopic constituents. These models are used to predict the behavior of systems that are inaccessible to experiment and obtain a fundamental understanding of the interactions that give rise to macroscopic observables. Such large-scale numerical calculations, utilizing high-performance computing, span a variety of molecular simulation techniques including both molecular dynamics (MD) and Monte Carlo (MD), as well as enhanced sampling biasing strategies such as well-tempered metadynamics, replica exchange, or umbrella sampling. The analysis and work-up of simulation trajectories encompasses numerical approaches to solving systems of differential equations, development of analytical theories to describe data, and software development.