Published March 25, 2026 | Version v0
Dataset Open

DESPOT: Direction-Enhanced Scoring POTentials

  • 1. ROR icon KU Leuven
  • 2. EDMO icon Katholieke Universiteit Leuven

Description

DESPOT: Direction-Enhanced Scoring POTentials

Overview

This repository contains the dataset accompanying the paper:

DESPOT: Direction-Enhanced Scoring POTentials 
Robin Poelmans, Bence Bruncsics, Adam Arany, Wout Van Eynde, Ahmed Shemy, Yves Moreau, Arnout RD Voet
bioRxiv 2026.03.31.714140; doi: https://doi.org/10.64898/2026.03.31.714140

This dataset contains the pre-trained DESPOT models, as well as the CASF-2016 benchmark set with all results.

Contents

| File | Description | Size |
|------|-------------|------|
| `atom_type_counts.csv` | Definition file for all internal atom types of DESPOT | 14.8 KB |
| `casf_pdb_ids.txt` | All 285 targets of CASF-2016 | 1.4 KB |
| `casf_2016.tar.gz` | CASF-2016 benchmark set. Subdirectory "benchmark-results" contains raw output for all tested models.  | 3.7 GB |
| `despot_counts_{dataset}.npz` | Raw interaction counts for DESPOT. 3 dataset options: {crown_train, crown_xtal, crown_leaky} | 20 MB |
| `despot_scores_{dataset}.npz` | Score matrices for DESPOT. 3 dataset options: {crown_train, crown_xtal, crown_leaky} | 800 MB |
| `despot_ds_scores_{dataset}.npz` | Score matrices for DESPOT-DS. 3 dataset options: {crown_train, crown_xtal, crown_leaky} | 800 KB |

Download and Installation

An automated download and installation pipeline is available in the accompanying GitHub repository:

https://github.com/KUL-LBMD/DESPOT

The pipeline handles downloading, integrity verification, and directory setup. Please refer to the GitHub repository for full instructions.

Quick start

git clone https://github.com/KUL-LBMD/DESPOT.git
cd DESPOT
# Follow the setup instructions in the GitHub README

License

This dataset is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

Citation

If you make use of DESPOT in your research, please cite our paper:

DESPOT: Direction-Enhanced Scoring POTentials 
Robin Poelmans, Bence Bruncsics, Adam Arany, Wout Van Eynde, Ahmed Shemy, Yves Moreau, Arnout RD Voet
bioRxiv 2026.03.31.714140; doi: https://doi.org/10.64898/2026.03.31.714140

Contact

For questions about the dataset, please open an issue on the GitHub repository or contact robin.poelmans@kuleuven.be.

Files

atom_type_counts.csv

Files (6.4 GB)

Name Size Download all
md5:6e55bab544fb69221c5603f617355643
14.3 kB Preview Download
md5:daa9cb4088844c28d179f3c1e5e5be3c
3.7 GB Download
md5:67f958b184c12862481268ba24fb080a
1.4 kB Preview Download
md5:9e0900a27fd254a065fbc7bc0f10aebe
23.9 MB Download
md5:375907b82a312b0740e02deddad5d7cb
20.6 MB Download
md5:be3c8e6e322511461c18bff5ac160c5b
20.7 MB Download
md5:eb9a909c611fe5a5bb911aebadb7f0a4
815.0 kB Download
md5:be3665875ee5a76ae4fba0c92e54c5d1
731.5 kB Download
md5:36ec48834625d5841ec698ec9ba7dd0c
833.3 kB Download
md5:529fd7056351929386ccadd6982b5873
898.0 MB Download
md5:ca5d00f3c0d57bd124ddbd96770f5bda
828.0 MB Download
md5:4d2d4f80d536001aee0e80a7a40e95c9
871.5 MB Download
md5:95a8abf906ce0e2122cbd633fae30a08
2.2 kB Preview Download

Additional details

Funding

Research Foundation - Flanders
AI-Driven Analysis and Design of Protein-Ligand Interactions for Advancing De Novo Drug Discovery 1176225N

Software

Repository URL
https://github.com/robin-poelmans/DESPOT
Programming language
Python
Development Status
Active