Geometric Deep Learning Reveals Ligandable and Cryptic RNA Binding Small Molecule Pockets (SMARTPocket)
Authors/Creators
- 1. University of Florida, Department of Medicinal Chemistry, Center for Natural Products, Drug Discovery and Development, Gainesville, FL 32610, USA
- 2. The Herbert Wertheim UF Scripps Institute for Biomedical Innovation and Technology, Department of Chemistry, 130 Scripps Way, Jupiter, FL 33458, USA
- 3. The Scripps Research Institute, Department of Chemistry, 130 Scripps Way, Jupiter, FL 33458, USA
- 4. University of Florida, Department of Computer & Information Science & Engineering, Gainesville, FL 32611, USA
Description
Overview
This is the official repository for SMARTPocket, an atomic-level geometric deep learning framework for predicting RNA-small molecule binding pockets directly from three-dimensional structure. Given an RNA structure in PDB format, SMARTPocket outputs a per-residue binding probability score.
Contents
This record provides the SMARTPocket codes (in the zip file), pre-trained model checkpoint and HDF5 datasets required to run inference, training, and evaluation.
- SMARTPocket.zip — SMARTPocket source code. Unzip to restore the full project. See the included README for setup and usage instructions.
- model_ckpt.pt — pre-trained SMARTPocket model checkpoint. Place at save/SMARTPocket/model_ckpt.pt.
- hariboss.h5 — HARIBOSS HDF5 dataset for training and evaluation. Place at data/h5_files/hariboss.h5.
- single_chain.h5 — single-chain benchmark HDF5 dataset. Place at data/h5_files/single_chain.h5.
Code and full documentation: https://github.com/AIDD-LiLab/SMARTPocket
Files
SMARTPocket.zip
Additional details
Related works
- Is supplement to
- Preprint: 10.64898/2026.06.18.732920 (DOI)