Related Experiment Video
Updated: May 18, 2026
07:19
Structural Studies of Macromolecules in Solution using Small Angle X-Ray Scattering
Published on: November 5, 2018
strainedSMILES2xyz: a workflow for reliable 3D structures of strained molecules from SMILES
Tori Demuth1, Julian Schnizer1, Dennis Svatunek2
1Institute of Applied Synthetic Chemistry, TU Wien, Getreidemarkt 9, 1060, Vienna, Austria.
Journal of Cheminformatics
|May 16, 2026
Summary
Generating accurate 3D molecular structures from SMILES is crucial for chemistry. Our new workflow, strainedSMILES2xyz, successfully generates correct 3D geometries for strained ring systems, outperforming existing methods.
Area of Science:
- Computational Chemistry
- cheminformatics
Background:
- Accurate 3D structure generation from SMILES is vital for data-driven chemistry.
- Existing methods often struggle with strained ring systems, leading to incorrect geometries.
Purpose of the Study:
- To develop an improved workflow for generating accurate 3D molecular structures from SMILES, specifically addressing challenges with strained ring systems.
- To provide a robust and accessible tool for computational chemists.
Main Methods:
- Introduced strainedSMILES2xyz, a Python workflow.
- Relaxed RDKit constraints to improve conformer generation.
- Explored stereoisomer variants and employed force-field refinement for error correction.
Main Results:
- Benchmarked strainedSMILES2xyz on both strained and unstrained ring systems.
- Demonstrated superior performance compared to existing tools.
- Achieved correct geometry generation in nearly all tested cases.
Conclusions:
- strainedSMILES2xyz significantly enhances the accuracy of 3D structure generation from SMILES, particularly for challenging strained ring systems.
- The workflow is available as a user-friendly Python package and Jupyter notebook.
