Related Experiment Video
Updated: Jun 28, 2025
11:20
Automated Protocols for Macromolecular Crystallization at the MRC Laboratory of Molecular Biology
Published on: January 24, 2018
16.4K
Automated BigSMILES conversion workflow and dataset for homopolymeric macromolecules
Sunho Choi1, Joonbum Lee2, Jangwon Seo1
1School of Electrical Engineering, Korea University, Seoul, South Korea.
Scientific Data
|April 11, 2024
Summary
This study introduces an automated workflow to convert Simplified Molecular-Input Line-Entry System (SMILES) into BigSMILES, enabling easier representation of macromolecules. This facilitates broader research and development applications in cheminformatics and artificial intelligence.
Area of Science:
- Cheminformatics
- Computational Chemistry
- Artificial Intelligence
Background:
- Simplified Molecular-Input Line-Entry System (SMILES) is widely used for chemical structure representation in AI.
- SMILES has limitations in representing complex macromolecules.
- BigSMILES was proposed as an alternative for macromolecules but requires significant preprocessing.
Purpose of the Study:
- To develop an automated conversion workflow from SMILES to BigSMILES for homopolymers.
- To provide a large dataset of BigSMILES representations for immediate research use.
- To validate the accuracy, interchangeability, and robustness of the generated BigSMILES.
Main Methods:
- Developed an automated conversion workflow.
- Generated BigSMILES representations for over 4.9 million homopolymer records.
- Implemented a rigorous validation process for the generated data.
- Documented utilized codes and functions for BigSMILES generation.
Main Results:
- Successfully generated 4,927,181 BigSMILES representations from SMILES.
- Validated the accuracy, interchangeability, and robustness of the BigSMILES conversion.
- Provided a comprehensive overview of the generation methodology.
Conclusions:
- The automated BigSMILES conversion workflow significantly aids researchers.
- This advancement facilitates further BigSMILES studies, including deep learning applications.
- The generated dataset enables immediate use in research and development for macromolecule representation.
