CMI researchers from Ames National Laboratory conducted the activity for this highlight
Innovation
Developed a cutting-edge machine learning (ML) model to accurately predict proton and metal-ligand binding constants for many metals and ligands.
Achievement
Trained on a diverse metal ions and ligands dataset, this model demonstrates greater generality, surpassing the performance of existing models limited to specific ligand and metal classes.
Significance and Impact
The remarkable performance using only Simplified Molecular-Input Line-Entry System (SMILES) strings as input allows accessibility to non-computational experts.
The next step will be to demonstrate the model's power by validating with new molecule design.
Hub Target Addressed
Highly selective extraction and separations. Increasing the speed of discovery and integration. Developing and applying scientific tools to accelerate technology maturation.