2020, Werner, Marco, Guo, Yachong, Baulin, Vladimir A.

Recent developments in computer processing power lead to new paradigms of how problems in many-body physics and especially polymer physics can be addressed. Parallel processors can be exploited to generate millions of molecular configurations in complex environments at a second, and concomitant free-energy landscapes can be estimated. Databases that are complete in terms of polymer sequences and architecture form a powerful training basis for cross-checking and verifying machine learning-based models. We employ an exhaustive enumeration of polymer sequence space to benchmark the prediction made by a neural network. In our example, we consider the translocation time of a copolymer through a lipid membrane as a function of its sequence of hydrophilic and hydrophobic units. First, we demonstrate that massively parallel Rosenbluth sampling for all possible sequences of a polymer allows for meaningful dynamic interpretation in terms of the mean first escape times through the membrane. Second, we train a multi-layer neural network on logarithmic translocation times and show by the reduction of the training set to a narrow window of translocation times that the neural network develops an internal representation of the physical rules for sequence-controlled diffusion barriers. Based on the narrow training set, the network result approximates the order of magnitude of translocation times in a window that is several orders of magnitude wider than the training window. We investigate how prediction accuracy depends on the distance of unexplored sequences from the training window. © 2020, The Author(s).

2019, Manzi, Berardo M., Werner, Marco, Ivanova, Elena P., Crawford, Russell J., Baulin, Vladimir A.

Recent technological advances have allowed the development of a new generation of nanostructured materials, such as those displaying both mechano-bactericidal activity and substrata that favor the growth of mammalian cells. Nanomaterials that come into contact with biological media such as blood first interact with proteins, hence understanding the process of adsorption of proteins onto these surfaces is highly important. The Random Sequential Adsorption (RSA) model for protein adsorption on flat surfaces was modified to account for nanostructured surfaces. Phenomena related to the nanofeature geometry have been revealed during the modelling process; e.g., convex geometries can lead to lower steric hindrance between particles, and hence higher degrees of surface coverage per unit area. These properties become more pronounced when a decrease in the size mismatch between the proteins and the surface nanostructures occurs. This model has been used to analyse the adsorption of human serum albumin (HSA) on a nano-structured black silicon (bSi) surface. This allowed the Blocking Function (the rate of adsorption) to be evaluated. The probability of the protein to adsorb as a function of the occupancy was also calculated.