Predicting Adhesive Free Energies of Polymer-Surface Interactions with Machine Learning

This repository contains an open source implementation of the machine learning model and corresponding dataset described in our paper Predicting Adhesive Free Energies of Polymer-Surface Interactions with Machine Learning.

Abstract

Polymer-surface interactions are crucial to many biological processes and industrial applications of polymers. Composition significantly influences a polymer's structural and functional properties, such as its solubility and accessible conformations. Here we propose a machine-learning method to connect a model polymer's composition with its adhesion to decorated surfaces. We simulate the adhesive free energies of 20,000 unique coarse-grained 1D sequential polymers interacting with functionalized surfaces and build support vector regression models that demonstrate inexpensive and reliable prediction of the adhesive free energy as a function of the sequence. Our work highlights the promising integration of coarse-grained simulation with data-driven machine learning methods for the design of new functional polymers and represents an important step toward linking polymer composition with polymer-surface interactions.

Dataset

The free energy is calculated by lammps and SSAGES in CG model with ABF sampling. You can check the simulation example.

Cite this work and star this repo

If this repository and dataset is helpful for your research please cite this work and star this repo.

Predicting Adhesive Free Energies of Polymer-Surface Interactions with Machine Learning Jiale Shi and Michael J. Quevillon and Pedro H. Amorim Valença and Jonathan K. Whitmer

@article{shi2022predicting,
author = {Shi, Jiale and Quevillon, Michael J. and Amorim Valença, Pedro H. and Whitmer, Jonathan K.},
title = {Predicting Adhesive Free Energies of Polymer–Surface Interactions with Machine Learning},
journal = {ACS Applied Materials \& Interfaces},
volume = {14},
number = {32},
pages = {37161-37169},
year = {2022},
doi = {10.1021/acsami.2c08891},
URL = {https://doi.org/10.1021/acsami.2c08891},
eprint = {https://doi.org/10.1021/acsami.2c08891}
}

Arxiv version is also available

Note

code and data for academic purpose only.