MD Simulations and Machine Learning to Quantify Interfacial Hydrophobicity

Explore the intersection of molecular dynamics simulations and machine learning to quantify interfacial hydrophobicity in this 32-minute conference talk. Delve into the challenges of analyzing high-dimensional datasets generated by classical molecular dynamics simulations and discover innovative machine learning approaches for efficient data analysis. Learn how these techniques can be applied to understand the hydrophobicity of functionalized interfaces, a crucial property in aqueous environments. Examine the impact of different data representations on machine learning method selection and prediction accuracy. Investigate the potential of topological data analysis in outperforming complex machine learning models when analyzing molecular dynamics output. Gain insights into the use of density matrices, hydrogen bond graphs, and Euler characteristics in hydrophobicity quantification. Understand the application of LASSO regression for feature selection and the analysis of important features in predicting interfacial hydrophobicity.

Intro
Combining Molecular Dynamics Simulations and Learning to Quantify Interfacial Hydrophobic
Motivation: Hydrophobicity of idealized and real interfaces
Hydration free energy of cavity as descriptor of hydrophobicity
Density matrices - orientation information
Density matrices-performance compariso
Hydrogen bond graphs
Topological data analysis
Euler Characteristic is stable
Human-selected water order parameters
Feature selection with LASSO regression
Analysis of important features
Acknowledgements