README.md

---
license: cc-by-4.0
language:
- en
tags:
- protein-ligand
- structural-biology
- drug-discovery
- PXR
- cofolding
- PoseBusters
pretty_name: Structure184 Five-Method PXR Cofolding Dataset
size_categories:
- 10K<n<100K
configs:
- config_name: models
  data_files:
  - split: models
    path: data/models.parquet
- config_name: posebusters_full
  data_files:
  - split: models
    path: data/posebusters_full.parquet
- config_name: ligands
  data_files:
  - split: ligands
    path: data/ligands.parquet
- config_name: truth_structures
  data_files:
  - split: structures
    path: data/truth_structures.parquet
---

# Structure184 five-method PXR cofolding dataset

This repository contains 92,000 PXR–ligand cofolded models:

- 184 ligands
- 20 seed positions
- 5 samples per seed
- Boltz2, Chai-1, ESMFold2, OpenFold3, and Protenix
- 18,400 models per method

Models use the governed prepared top solution-state ligand representation. Coordinates are harmonized to the 1NRL PXR frame using a 162-Cα core. The original generated coordinates are preserved up to that rigid alignment.

## Repository contents

| Path | Contents |
|---|---|
| `data/models.parquet` | Primary 92,000-row table: identifiers, confidence, structural metrics, OpenStructure scores, chemistry features, ProLIF, and compact PoseBusters status. |
| `data/posebusters_full.parquet` | Full PoseBusters 0.6.5 report for 91,734 topology-safe models. |
| `data/ligands.parquet` | One row per ligand and prepared state. |
| `data/truth_structures.parquet` | Reference-structure metadata. |
| `structures/<method>.tar.zst` | Aligned model CIFs, one archive per method. |
| `structures/truth_structures.tar.zst` | Aligned reference structures. |
| `COLUMN_GUIDE.md` | Concise column descriptions. |
| `MANIFEST.json` and `checksums.sha256` | Counts, sizes, and SHA-256 checksums. |

`models.parquet` has 119 columns. `structure_archive` and `structure_member` locate each CIF.

## Load the tables

```python
from datasets import load_dataset

models = load_dataset("dargason/structure184-five-method-cofolding", "models", split="models")
posebusters = load_dataset("dargason/structure184-five-method-cofolding", "posebusters_full", split="models")
```

Or with pandas:

```python
import pandas as pd

models = pd.read_parquet("hf://datasets/dargason/structure184-five-method-cofolding/data/models.parquet")
```

## Extract structures

```bash
tar --use-compress-program=unzstd -xf structures/boltz2.tar.zst
```

Archive members follow `method/ligand_id/model_id.cif`. Join them through `structure_member`.

## PoseBusters

PoseBusters was recomputed from scratch using generated ligand coordinates, authoritative prepared-state topology, and generated receptor coordinates.

- Computed: 91,734
- Explicit topology quarantine: 266
- Passed all configured checks: 75,668
- Failed one or more configured checks: 16,066

Quarantined rows remain in `models.parquet` with `posebusters_result_status=not_run_topology_quarantine`.

## Important interpretation notes

- PoseBusters is physical-validity QC, not pose-accuracy evidence.
- `protein_ligand_iptm` is the primary model-native confidence field; `iptm_source_key` records the engine-specific source.
- Columns beginning `official_ost_`, plus truth-relative RMSD and centroid-distance fields, are evaluation labels. Do not use them as blind model-selection inputs.
- ESMFold2 models are the campaign's no-MSA, 50-step protocol, not upstream-default ESMFold2.
- The public tables remove machine-local paths. File hashes and source-group identifiers are retained.
- This release does not claim that every engine parameter can be reconstructed from the public tables alone.

## Source and attribution

The ligand set and challenge context come from the [OpenADMET PXR Induction Blind Challenge](https://openadmet.ghost.io/announcing-the-next-openadmet-blind-challenge-predicting-pxr-induction/) and its [public challenge dataset](https://huggingface.co/datasets/openadmet/pxr-challenge-train-test). Please cite the OpenADMET challenge and this dataset repository when using these models or derived tables.

## Repository size

The packaged repository is approximately 4.6 GB compressed. The Parquet tables and `.tar.zst` structure archives are configured for Git LFS through `.gitattributes`.

## License

The tables, generated/aligned structures, truth-relative metrics, aligned truth structures, packaging, and documentation in this repository are provided under CC BY 4.0. Model-generating software remains under its respective license. See `LICENSE`.