---
license: other
tags:
  - chemistry
  - molecular-docking
  - virtual-screening
  - drug-discovery
  - gnina
  - dud-e
pretty_name: DUD-E Docked Poses (gnina virtual screening)
configs:
  - config_name: default
    data_files:
      - split: train
        path: dude_gnina_scores.parquet
---

# DUD-E Docked Poses (gnina virtual screening)

The docked poses from `dude_vs.tgz`, produced by the Koes lab at the University
of Pittsburgh for *Virtual Screening with Gnina 1.0*.

The original is one 14.7 GB archive that must be read start to finish. Here it is
split by target into `dude-vs/`, one archive per DUD-E target, so you can download
only the targets you need:

```bash
hf download Kingldore/dude-vs dude-vs/cdk2.tar.gz --repo-type dataset
tar -xzf dude-vs/cdk2.tar.gz
```

102 targets, 4,363 files, 14.7 GB in total. The contents are unchanged from the
original; only the packaging differs.

## Score table

`dude_gnina_scores.parquet` summarises every docked pose into one row per
`(target, compound_id)` — 1,422,897 rows, 22,784 actives and 1,400,113 decoys.

| column | meaning |
|---|---|
| `target` | DUD-E target name |
| `compound_id` | ZINC or CHEMBL identifier |
| `label` | 1 = active, 0 = decoy |
| `n_poses` | poses the statistics were computed over |
| `gnina_best_*` | value at the pose with the highest CNNscore |
| `gnina_mean_*`, `gnina_std_*` | mean and population standard deviation over all poses |

The four score families are `minimized_affinity`, `cnn_score`, `cnn_affinity`,
and `cnn_vs` (CNNscore x CNNaffinity, computed per pose then aggregated).

`minimized_affinity` is the empirical docking score, from the column the summary
files label `Vina`. DUD-E was docked with default smina arguments
(`--seed 0 --autobox_add 4 --num_modes 9`), so this is the Vina scoring
function, and `num_modes 9` is why most ligands have nine poses. The pose files
are named `*_docked_vinardo.sdf.gz` because Vinardo rescoring was also carried
out; those scores are in a separate `vinardo.summary` and are not included here.

All four `gnina_best_*` values come from the **same** pose — the one with the
highest CNNscore — rather than each column being maximised independently.

Most ligands have 9 poses. A few are docked more than once against a target;
identical pose rows are dropped and the rest pooled, so `n_poses` can exceed 9.

**Original download:** <https://bits.csb.pitt.edu/files/gninavs/dude_vs.tgz>

## Citation

Please cite both the method and the underlying benchmark.

**gnina virtual screening** — the source of these docked poses:

```bibtex
@article{sunseri2021virtual,
  title   = {Virtual Screening with Gnina 1.0},
  author  = {Sunseri, Jocelyn and Koes, David Ryan},
  journal = {Molecules},
  volume  = {26},
  number  = {23},
  pages   = {7369},
  year    = {2021},
  doi     = {10.3390/molecules26237369}
}
```

**DUD-E** — the benchmark set that was docked:

```bibtex
@article{mysinger2012directory,
  title   = {Directory of Useful Decoys, Enhanced (DUD-E): Better Ligands and
             Decoys for Better Benchmarking},
  author  = {Mysinger, Michael M. and Carchia, Michael and Irwin, John J. and
             Shoichet, Brian K.},
  journal = {Journal of Medicinal Chemistry},
  volume  = {55},
  number  = {14},
  pages   = {6582--6594},
  year    = {2012},
  doi     = {10.1021/jm300687e}
}
```

## Links

- Paper: <https://doi.org/10.3390/molecules26237369>
- DUD-E paper: <https://doi.org/10.1021/jm300687e>
- gnina software: <https://github.com/gnina/gnina>
- DUD-E: <https://dude.docking.org/>
- Koes lab file index: <https://bits.csb.pitt.edu/files/>

## License

This is a mirror hosted for convenience; the data belongs to its original
authors and no additional licence is granted here.