Diamond Energy III

Machine Learning Accelerated Diamond Energy Conformer Search Project

Mengman Wei, Jonathan Goodman

In previous studies, it was discovered that local minima with structures resembling diamond lattices can significantly expedite the conformer search process. This project aims to utilize low-energy conformers obtained from earlier research(https://github.com/Goodman-lab/Diamond_energy and https://github.com/Goodman-lab/Diamond-Energy-II). Specifically, we will:

1. Investigate whether current leading ML-based conformer search algorithms can rapidly learn these diamond lattice-like low-energy conformer spaces.
2. Test whether well-trained models, based on these low-energy spaces, can effectively guide a quick conformer search.

Contents

1. Diamond Energy Projects Low Energy Conformers Compact Data

   • Detailed information and datasets of low-energy conformers derived from previous studies(https://github.com/Goodman-lab/Diamond_energy and https://github.com/Goodman-lab/Diamond-Energy-II).
   • These data could be access through this link: https://drive.google.com/file/d/1A3gPoIelP-tkX3-SpkcyD6CSiZfMUAq8/view?usp=drive_link

2. Scripts Used for Data Preparation and Training

   • Collection of scripts and tools for processing and preparing data for machine learning training.

3. Scripts for Post-Training Data Analysis

   • Scripts to handle and analyze data obtained from well-trained ML models.

4. Results Summary

   • Comprehensive summary of findings and insights.

5. All Calculations Rawdata

   • All raw data from the calculations conducted in this project could be accessed from this link(https://drive.google.com/drive/folders/1-vs6JgZgpJkgCfYGLK1gGT6BTo0br3Zz?usp=drive_link).

Reference

@phdthesis{Wei2024DiamondEnergy,
  title={*Diamond Energy* – a systematic conformation searching method},
  author={Wei, Mengman},
  year={2024},
  school={University of Cambridge}
  doi={https://doi.org/10.17863/CAM.109200}
}