-
This repository presents a novel computational framework that integrates the biophysics-based Peptide Binder Design (PepBD) algorithm with evidential deep learning (EDL) and metaheuristic search to accelerate the design of plastic-binding peptides.
-
The framework leverages the predictive power of EDL to efficiently explore the vast peptide sequence space, while incorporating uncertainty quantification to prioritize promising candidates.
-
Molecular dynamics simulations validate the efficacy of the designed peptides, demonstrating their superior binding affinity compared to existing solutions, thus opening a new avenue for bio-inspired microplastic remediation strategies.
@article{https://doi.org/10.1002/aic.17469,
author = {Abdulelah S. Alshehri, Michael T. Bergman, Carol K. Hall, and Fengqi You},
title = {Biophysics-Informed Evidential Deep Learning for Uncertainty-Aware Design of Plastic-Binding Peptides},
journal = {},
volume = {},
number = {},
pages = {},
keywords = {},
doi = {},
url = {},
eprint = {,
abstract = {}
1 Data/contains PepBD data results2 Models/contains EDL moddels3 Codes/contains peptide representation and genetic algorithm generation methods4 Results/contains the results of EDL and molecular dynamic simulations along with molecular dynamics comparisons between PepBD and EDL
- python (>=3.7)
- tensorflow (>=2.0)
- evidential-deep-learning
- Pickle
- functools
- pymoo
To install evidential-deep-learning:
pip install evidential-deep-learning
To install pymoo:
git clone https://github.com/anyoptimization/pymoo
cd pymoo
pip install .