A fully automated, cloud-based crystal structure prediction protocol using a purpose-built neural network potential claims to generate and correctly rank all 110 Z'=1 polymorphs across 49 drug-like molecules at roughly 8.4k CPU hours per molecule.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
physics.chem-ph 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Toward Routine CSP of Pharmaceuticals: A Fully Automated Protocol Using Neural Network Potentials
A fully automated, cloud-based crystal structure prediction protocol using a purpose-built neural network potential claims to generate and correctly rank all 110 Z'=1 polymorphs across 49 drug-like molecules at roughly 8.4k CPU hours per molecule.