All-atom molecular dynamics simulations hold a great promise as a valuable tool for understanding and predicting protein-surface interactions. However, experimental data is required in order to tune and validate these computational methods before they can be utilized with confidence. The main difficulty in direct comparison of experiments with simulations, however, is that experiments provide averaged properties of billions of adsorbed proteins over timeframes of seconds while conventional MD gives results for a single adsorbed protein over tens of nanoseconds. To address this issue, we use an advanced sampling algorithm, TIGER2, to generate an equilibrated Boltzmann-weighted ensemble of states of a molecular system from which equilibrated ensemble-average properties can be calculated for direct comparison with experiment. To determine the most stable configurations of the system, the ensembles of sampled states must be analyzed through cluster analysis. Cluster analysis methods for analyzing MD trajectories have traditionally been developed for the analysis of protein structure in solution, but not for an ensemble of proteins in their adsorbed state. The objective of this research was, therefore, to develop and apply a cluster analysis method that can be used for the analysis of a large ensemble of states obtained from a simulation of protein adsorption behavior. The application of cluster analysis in the protein-adsorption simulations requires a different approach from regular protein-in-water simulations to structurally align adsorbed protein. For adsorbed-protein simulation, it is important to compare both tertiary conformation of the adsorbed state of the protein and its orientation over the material surface while tertiary conformational comparison alone is required for the analysis of solution behavior. We evaluate clustering analysis methods that addressed both orientation and conformation. Several well-known clustering algorithms, including hierarchical and partitional clustering techniques were examined to establish and validate an algorithm that was well-suited for the analysis of adsorbed protein configurations.

Cluster analysis of molecular dynamics trajectories in protein adsorption simulations

247th National Meeting (2014)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.