Dataset Viewer
Auto-converted to Parquet Duplicate
text
stringlengths
19
19
AF-0000000065760031
AF-0000000065760112
AF-0000000065760132
AF-0000000065760399
AF-0000000065760503
AF-0000000065760505
AF-0000000065760511
AF-0000000065760556
AF-0000000065760566
AF-0000000065760657
AF-0000000065760768
AF-0000000065760794
AF-0000000065760829
AF-0000000065760832
AF-0000000065760844
AF-0000000065760935
AF-0000000065761048
AF-0000000065761064
AF-0000000065761100
AF-0000000065761110
AF-0000000065761113
AF-0000000065761114
AF-0000000065761115
AF-0000000065761117
AF-0000000065761122
AF-0000000065761145
AF-0000000065761226
AF-0000000065761387
AF-0000000065761514
AF-0000000065761518
AF-0000000065761520
AF-0000000065761526
AF-0000000065761537
AF-0000000065761789
AF-0000000065761793
AF-0000000065761911
AF-0000000065761914
AF-0000000065761933
AF-0000000065761977
AF-0000000065761978
AF-0000000065761982
AF-0000000065761984
AF-0000000065762011
AF-0000000065762016
AF-0000000065762021
AF-0000000065762024
AF-0000000065762026
AF-0000000065762028
AF-0000000065762042
AF-0000000065762092
AF-0000000065762098
AF-0000000065762099
AF-0000000065762340
AF-0000000065762458
AF-0000000065762551
AF-0000000065762598
AF-0000000065763035
AF-0000000065763043
AF-0000000065763146
AF-0000000065763147
AF-0000000065763150
AF-0000000065763157
AF-0000000065763160
AF-0000000065763161
AF-0000000065763176
AF-0000000065763179
AF-0000000065763181
AF-0000000065763185
AF-0000000065763186
AF-0000000065763244
AF-0000000065763254
AF-0000000065763386
AF-0000000065763493
AF-0000000065763509
AF-0000000065763514
AF-0000000065763519
AF-0000000065763534
AF-0000000065763536
AF-0000000065763565
AF-0000000065763663
AF-0000000065764062
AF-0000000065764064
AF-0000000065764065
AF-0000000065764072
AF-0000000065764073
AF-0000000065764079
AF-0000000065764088
AF-0000000065764089
AF-0000000065764308
AF-0000000065764322
AF-0000000065764395
AF-0000000065764439
AF-0000000065764446
AF-0000000065764496
AF-0000000065764527
AF-0000000065764564
AF-0000000065764567
AF-0000000065764790
AF-0000000065764842
AF-0000000065764877
End of preview. Expand in Data Studio

K-Fold Structure Benchmark (AFDB-Multimer)

Everything needed to run the folding / co-folding / probing benchmarks except per-encoder features (which you extract from k-fold-structure/repsp-triprorep-tokens with your encoder, see the GitHub release README).

The benchmark goes apo → holo: predict properties of the bound homodimer ("holo") from the unbound monomer ("apo"). The two chains are identical (homodimer), so each residue position carries a single label.

Contents

splits/
    folding/{train,valid,test}.txt        # 390,627 / 400 / 1,000 PIDs (cluster reps, LMDB-cleaned)
    probing/splits_{train,valid,test}.txt # 39,076 / 400 / 1,000 PIDs (10% of folding train)
probing/
    labels.pkl                            # per-residue labels: binding_site / delta_sasa_mean / levy_tier / bond_type_plip
boltz_apo_tokens/   shard*.tar            # Boltz-tokenized apo monomers (folding input)
boltz_apo_targets/  shard*.tar            # apo structure targets
boltz_holo_tokens/  shard*.tar            # Boltz-tokenized holo dimers (co-folding input)
boltz_holo_targets/ shard*.tar            # holo structure targets

Each boltz_* dir is split into ~5 GB tar shards (the native dirs hold ~395k small files each, too many to upload individually). Extract all shards of a dir into one folder:

hf download k-fold-structure/repsp-benchmark --repo-type dataset \
    --include "boltz_holo_tokens/*" --local-dir ./benchmark
cd ./benchmark/boltz_holo_tokens
for t in shard*.tar; do tar xf "$t"; done && rm shard*.tar
# -> manifest.json  records/*.json  tokens/*.pkl

Note on "Boltz": the boltz_* dirs use the tokenized layout the folding trunk (SimpleFold) reads, manifest.json + records/ + tokens/ (or structures/ for targets). After extracting, point the folding datamodule's tokenized_dir / target_dir at these folders.

Notes

  • Splits are LMDB-cleaned: ~0.06% of train PIDs that failed tokenization are already removed, so you won't hit KeyErrors.
  • The folding symmetries config field is unused for this protein-only benchmark (no ligands, chain A<->B symmetry is computed on-the-fly).

License

MIT

Downloads last month
204