Interaction Profiles as a Universal Language for Generative Molecular Design with ShEPhERD-2
- 1Massachusetts Institute of Technology
Abstract
Three-dimensional intermolecular interactions govern molecular recognition and are fundamental to the pharmacological activity of small-molecule drugs. We propose that an interaction profile, comprising shape, electrostatics, and directional pharmacophores, is a sufficient and transferable design specification for molecular design. We introduce ShEPhERD-2, a 3D generative model that generates molecular structures conditioned on explicit interaction profiles. ShEPhERD-2 generates low-strain, drug-like molecules more efficiently and with greater fidelity to target interaction profiles than its predecessor while enabling precise control over molecular design through pharmacophore prioritization, substructure constraints, and composition of multiple interaction profiles. We demonstrate the application of a single model to bioisosteric fragment merging, dual-target design, selectivity engineering, and modality hopping without task-specific retraining. These results establish interaction profiles as a chemotype-agnostic interface between structure hypotheses and molecular design.
Related
- Generative design of bacteriophages with genome language models
- A deep generative decoder predicts microRNA expression from bulk and single-cell mRNA profiles
- Generative Design of New-to-nature Biosynthetic Assembly Lines with Genomic Language Modeling
- PepRePs: Peptide-Retargeted Phosphatases Designed with Generative Language Models
- Hi-cGAN: Prediction of Hi-C interaction matrices with conditional generative adversarial networks
BibTeX
@misc{abeywardane2026interaction,
title = {{Interaction Profiles as a Universal Language for Generative Molecular Design with ShEPhERD-2}},
author = {K. A. Abeywardane and K. Walker and C. W. Coley},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.10.750648},
url = {https://www.biorxiv.org/content/10.64898/2026.09.10.750648v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Interaction Profiles as a Universal Language for Generative Molecular Design with ShEPhERD-2
AU - K. A. Abeywardane
AU - K. Walker
AU - C. W. Coley
PY - 2026
DA - 2026/09/10
PB - bioRxiv
DO - 10.64898/2026.09.10.750648
UR - https://www.biorxiv.org/content/10.64898/2026.09.10.750648v1
AB - Three-dimensional intermolecular interactions govern molecular recognition and are fundamental to the pharmacological activity of small-molecule drugs. We propose that an interaction profile, comprising shape, electrostatics, and directional pharmacophores, is a sufficient and transferable design specification for molecular design. We introduce ShEPhERD-2, a 3D generative model that generates molecular structures conditioned on explicit interaction profiles. ShEPhERD-2 generates low-strain, drug-like molecules more efficiently and with greater fidelity to target interaction profiles than its predecessor while enabling precise control over molecular design through pharmacophore prioritization, substructure constraints, and composition of multiple interaction profiles. We demonstrate the application of a single model to bioisosteric fragment merging, dual-target design, selectivity engineering, and modality hopping without task-specific retraining. These results establish interaction profiles as a chemotype-agnostic interface between structure hypotheses and molecular design.
ER -