sammargolis/drugtargetbench-assets

DrugTargetBench assets

Immutable per-world payloads for the DrugTargetBench Harbor benchmark. You do not download this repository by hand: Harbor resolves the archives a task needs and verifies them against the frozen manifest before the agent starts.

uv tool install harbor
harbor run -d drugtargetbench/drugtargetbench@v1.0 -a <agent> -m <model>

Layout

worlds/<world-id>/tables.tar            omics, tabular, data dictionary
worlds/<world-id>/signals.tar           ECG waveforms, coronary angiograms
worlds/<world-id>/oracle-state.tar      experiment-service state
worlds/<world-id>/imaging-NNN.tar       cine-MRI, ~4 GB shards
worlds/<world-id>/imaging-visit2-NNN.tar
worlds/<world-id>/metadata.json
manifests/assets.json
manifests/sha256sums.txt

Twenty worlds, one independently addressable payload each. The three Harbor budget regimes for a world reference the same payload, so a full 60-task sweep materialises 345 GB once rather than 1.03 TB.

Publication status

Every archive carries a status in manifests/assets.json:

The cine-MRI is derived from the ACDC dataset, which is registration-gated and may not be redistributed. Those archives are listed in the manifest with final hashes but are served only where redistribution rights permit. Where they are absent, Harbor rebuilds them from an authorised local copy:

export DTB_ACDC_ROOT=/path/to/authorized/ACDC/training

The resulting files are verified against the same SHA-256 values, so a trial is byte-identical whether the imaging was downloaded or rendered locally.

Integrity

manifests/sha256sums.txt covers every archive. Harbor checks each one after download and refuses to run on a mismatch rather than scoring a partial world. Always pin an immutable commit sha; never run against main.

Citation

Cite DrugTargetBench, and cite ACDC separately if you use the real-anatomy imaging: Bernard et al., Deep Learning Techniques for Automatic MRI Cardiac Multi-Structures Segmentation and Diagnosis, IEEE TMI 2018.