Data-Driven Prediction of Protein Adsorption on Self-Assembled Monolayers toward Material Screening and Design
Rudolf Jason Kwaria, Evan Angelo Quimada Mondarte, Hiroyuki Tahara, Ryongsok Chang, T. Hayashi
Abstract
We attempt to predict the water contact angle (WCA) of self-assembled monolayers (SAMs) and protein adsorption on the SAMs from the chemical structures of molecules constituting the SAMs using machine learning with an artificial neural network (ANN) model. After training the ANN with data of 145 SAMs, the ANN became capable of predicting the WCA and protein adsorption accurately. The analysis of the trained ANN quantitatively revealed the importance of each structural parameter for the WCA and protein adsorption, providing essential and quantitative information for material design. We found that the degree of importance agrees well with our general perception on the physicochemical properties of SAMs. We also present the prediction of the WCA and protein adsorption of hypothetical SAMs and discuss the possibility of our approach for the material screening and design of SAMs with desired functions. On the basis of these results, we also discuss the limitation of this approach and prospects.