Dataset Viewer
Auto-converted to Parquet Duplicate
target_id
stringlengths
15
15
disease_id
stringlengths
10
15
transition_year
int64
2k
2.02k
outcome
bool
2 classes
ENSG00000177455
EFO_0000574
2,014
false
ENSG00000197170
EFO_0000565
2,015
false
ENSG00000169083
EFO_0003869
2,015
false
ENSG00000131495
EFO_0000673
2,011
false
ENSG00000186716
EFO_0005221
2,015
false
ENSG00000065361
EFO_0003060
2,002
true
ENSG00000128059
EFO_0005676
2,007
false
ENSG00000176890
EFO_0002916
2,000
true
ENSG00000164182
EFO_0000616
2,012
false
ENSG00000103546
EFO_0000384
2,005
false
ENSG00000152270
EFO_0002970
2,012
false
ENSG00000061273
MONDO_0000430
2,013
false
ENSG00000257365
MONDO_0019469
2,006
false
ENSG00000175482
EFO_0000211
2,001
true
ENSG00000186350
EFO_0000574
2,001
true
ENSG00000140443
MONDO_0005575
2,007
false
ENSG00000043591
MONDO_0011382
2,010
false
ENSG00000232810
MONDO_0017816
2,001
false
ENSG00000146143
MONDO_0021063
2,004
false
ENSG00000125356
EFO_0000764
2,015
false
ENSG00000151834
EFO_0004262
1,997
true
ENSG00000170425
EFO_0009492
2,001
false
ENSG00000096384
MONDO_0001056
2,010
false
ENSG00000070886
EFO_0003869
2,002
false
ENSG00000188641
MONDO_0007576
2,008
true
ENSG00000156738
EFO_1000749
2,003
true
ENSG00000274286
EFO_0003768
1,998
false
ENSG00000131747
MONDO_0002974
2,001
false
ENSG00000136521
EFO_0000196
2,013
false
ENSG00000165995
MONDO_0004986
2,015
false
ENSG00000152086
EFO_0000756
2,002
true
ENSG00000123416
MONDO_0013730
2,012
false
ENSG00000185313
HP_0011868
2,001
false
ENSG00000151834
HP_0011868
2,001
false
ENSG00000101680
EFO_0009085
2,005
false
ENSG00000196591
EFO_0006861
2,005
false
ENSG00000198695
HP_0000855
2,004
true
ENSG00000113448
EFO_0003956
2,006
false
ENSG00000061918
EFO_0000266
2,015
false
ENSG00000171557
MONDO_0009009
2,013
false
ENSG00000187498
MONDO_0008231
2,008
true
ENSG00000131398
MONDO_0043797
2,004
false
ENSG00000140443
EFO_1001480
2,007
false
ENSG00000131759
EFO_1000726
2,012
false
ENSG00000183454
MONDO_0005277
2,014
false
ENSG00000125691
MONDO_0009061
2,005
true
ENSG00000145494
Orphanet_309005
2,006
true
ENSG00000198056
MONDO_0017198
2,007
false
ENSG00000137285
EFO_1000044
2,013
true
ENSG00000165646
MONDO_0005090
2,009
true
ENSG00000100479
MONDO_0000430
2,013
false
ENSG00000082701
Orphanet_98757
2,011
false
ENSG00000170425
HP_0001643
2,008
false
ENSG00000171873
MONDO_0005301
2,007
false
ENSG00000160194
EFO_1001459
2,012
false
ENSG00000185527
HP_0001643
2,008
false
ENSG00000142627
EFO_0003841
2,007
false
ENSG00000073756
EFO_0000389
2,007
false
ENSG00000073756
EFO_1000843
2,015
false
ENSG00000167553
EFO_0000685
2,002
false
ENSG00000120659
EFO_0003869
2,013
false
ENSG00000136160
EFO_0005207
2,010
true
ENSG00000261456
HP_0100543
2,006
false
ENSG00000167553
EFO_1001094
2,014
false
ENSG00000153253
EFO_0006911
2,011
false
ENSG00000101868
EFO_1000613
2,005
false
ENSG00000198763
Orphanet_309005
2,006
true
ENSG00000103546
EFO_0009687
2,012
false
ENSG00000167553
MONDO_0001187
1,996
true
ENSG00000160716
EFO_0004888
2,007
false
ENSG00000171848
MONDO_0003060
2,002
true
ENSG00000095303
EFO_0003764
2,008
false
ENSG00000139842
EFO_0000403
2,007
true
ENSG00000268089
EFO_0003756
2,010
false
ENSG00000105723
EFO_1001050
2,009
false
ENSG00000165731
MONDO_0005575
2,007
true
ENSG00000037280
MONDO_0018364
2,007
false
ENSG00000004779
EFO_0000616
2,012
false
ENSG00000149305
MONDO_0004985
2,004
false
ENSG00000176014
EFO_0003833
1,998
false
ENSG00000198216
EFO_0000512
2,013
false
ENSG00000172572
EFO_0000220
2,015
false
ENSG00000170425
HP_0001397
2,013
false
ENSG00000137806
MONDO_0007254
2,009
true
ENSG00000164400
MONDO_0004979
2,012
false
ENSG00000134640
HP_0000618
2,006
false
ENSG00000128039
MONDO_0008315
2,004
true
ENSG00000198763
EFO_0002950
2,010
true
ENSG00000137673
EFO_1001459
2,009
false
ENSG00000159352
MONDO_0018906
2,013
false
ENSG00000090266
EFO_0003769
2,014
false
ENSG00000151366
EFO_0003940
2,015
false
ENSG00000117461
MONDO_0002108
2,013
false
ENSG00000061273
EFO_0000519
2,010
false
ENSG00000113721
EFO_0003897
2,006
false
ENSG00000118523
EFO_0004239
2,010
false
ENSG00000163285
EFO_0009686
2,002
true
ENSG00000261456
EFO_1000403
2,008
false
ENSG00000100519
EFO_1001469
2,003
true
ENSG00000170906
EFO_0002617
2,011
false
End of preview. Expand in Data Studio

THBKG — Temporal Heterogeneous Biomedical Knowledge Graph

A dated biomedical knowledge graph built from Open Targets 26.03 (with Reactome, ChEMBL and ClinicalTrials.gov), plus a clinical-advancement benchmark: rank target–disease pairs by their likelihood of advancing to Phase II, scored only from evidence datable strictly before each pair's decision year.

Every temporal edge carries the year its evidence first appeared, so the graph can be queried as of any historical decision point without leakage.

Files

Path What it is
graph.safetensors All graph tensors (node features, edge_index, edge_attr, edge_time), pickle-free.
metadata.json Schema to reassemble the graph: node/edge types, tensor keys, shapes, dtypes.
mappings/<ntype>.parquet node_id ↔ index for each node type (target, disease, molecule, go, reactome).
labels/train.parquet, labels/eval.parquet Advancement labels: target_id, disease_id, transition_year, outcome.
load_thbkg.py Standalone loader — rebuilds a PyG HeteroData, reads the label splits.
croissant.json MLCommons Croissant metadata.

Graph at a glance

  • Nodes: 110,396 across five types — target, disease, molecule, Gene Ontology term (go), Reactome pathway (reactome). Feature dims: molecule 1024, disease 256, target 56, GO 64, Reactome 64.
  • Edges: ~11.1M over nineteen relations (sixteen temporal, three static); each temporal edge carries edge_time (year first observed) and edge_attr = [weight, novelty].
  • Supervision is external. The graph holds only nodes and evidence edges; the clinical-advancement outcomes live in the label parquets and are joined to the graph by node ID at train/eval time. (Baking the label in as an edge made supervision indistinguishable from evidence and leaked it into message passing.)

Advancement benchmark

28,795 target–disease pairs, split temporally: 21,602 train (Phase II entry 1995–2015) / 7,193 eval (2016–2021, ~9.3% positive). outcome is True iff a Phase III trial for the same pair was registered within three years of Phase II entry. Each pair is scored using only graph evidence dated strictly before its transition_year.

Primary metric: Relative Success @ K (an importance-weighted hit rate), reported per therapeutic area and Wilcoxon-tested against a randomized-decisions baseline.

Quick start

from huggingface_hub import snapshot_download
from load_thbkg import load_graph, load_labels, node_index

path  = snapshot_download("<user>/THBKG", repo_type="dataset")
data  = load_graph(path)              # PyG HeteroData (reverse edges added, training view)
train = load_labels(path, "train")    # DataFrame: target_id, disease_id, transition_year, outcome
eval_ = load_labels(path, "eval")

# resolve a label endpoint to its graph node index
t2i, d2i = node_index(path, "target"), node_index(path, "disease")

The label parquets also load directly as a 🤗 datasets config:

from datasets import load_dataset
ds = load_dataset("<user>/THBKG", "advancement")   # splits: train, eval

License & citation

Dataset artifacts: CC-BY-4.0. Source code (construction pipeline, benchmark harness): MIT, at https://github.com/jackysiupuichung/THBKG. Archival DOI on Zenodo: concept 10.5281/zenodo.20795231.

Siu, Cabrera, Mudaliar, and Zubiaga.
THBKG: A Temporal Biomedical Knowledge Graph for Leak-Free Clinical Advancement Prediction.
Downloads last month
24