Datasets:
Formats:
parquet
Languages:
English
Size:
10M - 100M
Tags:
biology
chemistry
drug-discovery
clinical-trials
protein-protein-interaction
gene-essentiality
License:
Add data exports: LLM benchmarks, CT, PPI small files
Browse files- data/README.md +123 -0
- data/compound_names.parquet +3 -0
- data/croissant.json +176 -0
- data/ct/negbiodb_ct_m1_balanced.parquet +3 -0
- data/ct/negbiodb_ct_m1_degree_matched.parquet +3 -0
- data/ct/negbiodb_ct_m1_realistic.parquet +3 -0
- data/ct/negbiodb_ct_m1_smiles_only.parquet +3 -0
- data/ct/negbiodb_ct_m1_uniform_random.parquet +3 -0
- data/ct/negbiodb_ct_m2.parquet +3 -0
- data/ct/negbiodb_ct_pairs.parquet +3 -0
- data/ct_llm/ct_l1_dataset.jsonl +0 -0
- data/ct_llm/ct_l1_metadata.json +22 -0
- data/ct_llm/ct_l2_dataset.jsonl +0 -0
- data/ct_llm/ct_l2_metadata.json +24 -0
- data/ct_llm/ct_l3_dataset.jsonl +0 -0
- data/ct_llm/ct_l3_metadata.json +13 -0
- data/ct_llm/ct_l4_dataset.jsonl +0 -0
- data/ct_llm/ct_l4_metadata.json +19 -0
- data/ge_llm/ge-l1_metadata.json +22 -0
- data/ge_llm/ge-l2_metadata.json +17 -0
- data/ge_llm/ge-l3_metadata.json +15 -0
- data/ge_llm/ge-l4_metadata.json +17 -0
- data/ge_llm/ge_l1_dataset.jsonl +0 -0
- data/ge_llm/ge_l2_dataset.jsonl +0 -0
- data/ge_llm/ge_l3_dataset.jsonl +0 -0
- data/ge_llm/ge_l4_dataset.jsonl +0 -0
- data/llm_benchmarks/l1_mcq.jsonl +0 -0
- data/llm_benchmarks/l2_candidates.jsonl +0 -0
- data/llm_benchmarks/l3_reasoning_pilot.jsonl +50 -0
- data/llm_benchmarks/l4_tested_untested.jsonl +0 -0
- data/ppi_llm/ppi_l1_dataset.jsonl +0 -0
- data/ppi_llm/ppi_l1_metadata.json +22 -0
- data/ppi_llm/ppi_l2_dataset.jsonl +0 -0
- data/ppi_llm/ppi_l2_metadata.json +17 -0
- data/ppi_llm/ppi_l3_dataset.jsonl +0 -0
- data/ppi_llm/ppi_l3_metadata.json +19 -0
- data/ppi_llm/ppi_l4_dataset.jsonl +0 -0
- data/ppi_llm/ppi_l4_metadata.json +17 -0
data/README.md
ADDED
|
@@ -0,0 +1,123 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-sa-4.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
tags:
|
| 6 |
+
- drug-target-interaction
|
| 7 |
+
- clinical-trials
|
| 8 |
+
- protein-protein-interaction
|
| 9 |
+
- negative-results
|
| 10 |
+
- benchmark
|
| 11 |
+
- bioinformatics
|
| 12 |
+
task_categories:
|
| 13 |
+
- text-classification
|
| 14 |
+
- question-answering
|
| 15 |
+
- text-generation
|
| 16 |
+
size_categories:
|
| 17 |
+
- 10M<n<100M
|
| 18 |
+
---
|
| 19 |
+
|
| 20 |
+
# NegBioDB: A Negative Results Database for Biomedical Sciences
|
| 21 |
+
|
| 22 |
+
NegBioDB is a large-scale database of experimentally confirmed negative results across three biomedical domains, paired with dual ML/LLM benchmarks for evaluating how well computational methods handle negative evidence.
|
| 23 |
+
|
| 24 |
+
## Overview
|
| 25 |
+
|
| 26 |
+
| Domain | Negative Results | Entities | Benchmark Tasks |
|
| 27 |
+
|--------|-----------------|----------|-----------------|
|
| 28 |
+
| DTI (Drug-Target Interaction) | 24,965,618 pairs | 919K compounds, 3.7K targets | ML (M1) + LLM (L1-L4) |
|
| 29 |
+
| CT (Clinical Trial Failure) | 102,850 results | 216K trials, 176K interventions | ML (M1-M2) + LLM (L1-L4) |
|
| 30 |
+
| PPI (Protein-Protein Interaction) | 2,220,786 pairs | 18.4K proteins | ML (M1) |
|
| 31 |
+
|
| 32 |
+
## Data Sources
|
| 33 |
+
|
| 34 |
+
### DTI Domain
|
| 35 |
+
| Source | License | Contribution |
|
| 36 |
+
|--------|---------|-------------|
|
| 37 |
+
| ChEMBL 34 | CC BY-SA 3.0 | Curated bioactivity data |
|
| 38 |
+
| PubChem BioAssay | Public domain | HTS screening results |
|
| 39 |
+
| BindingDB | CC BY-SA 3.0 | Binding measurements |
|
| 40 |
+
| DAVIS | CC BY 4.0 | Kinase selectivity panel |
|
| 41 |
+
|
| 42 |
+
### CT Domain
|
| 43 |
+
| Source | License | Contribution |
|
| 44 |
+
|--------|---------|-------------|
|
| 45 |
+
| AACT (ClinicalTrials.gov) | Public domain | Trial metadata |
|
| 46 |
+
| Open Targets | Apache 2.0 | Drug-target mappings |
|
| 47 |
+
| CTO | MIT | Clinical trial outcomes |
|
| 48 |
+
| Shi & Du 2024 | CC BY 4.0 | Safety/efficacy data |
|
| 49 |
+
|
| 50 |
+
### PPI Domain
|
| 51 |
+
| Source | License | Contribution |
|
| 52 |
+
|--------|---------|-------------|
|
| 53 |
+
| IntAct | CC BY 4.0 | Curated non-interactions |
|
| 54 |
+
| HuRI | CC BY 4.0 | Y2H systematic negatives |
|
| 55 |
+
| hu.MAP 3.0 | CC BY 4.0 | Complex-derived negatives |
|
| 56 |
+
| STRING v12.0 | CC BY 4.0 | Zero-score protein pairs |
|
| 57 |
+
|
| 58 |
+
## File Structure
|
| 59 |
+
|
| 60 |
+
### DTI Files
|
| 61 |
+
| File | Size | Rows | Description |
|
| 62 |
+
|------|------|------|-------------|
|
| 63 |
+
| `negbiodb_dti_pairs.parquet` | 139.4 MB | 24,965,618 | All negative DTI pairs with metadata |
|
| 64 |
+
| `negbiodb_m1_balanced.parquet` | 270.2 MB | 1,725,446 | M1 balanced dataset (1:1 pos:neg) |
|
| 65 |
+
| `negbiodb_m1_realistic.parquet` | 753.3 MB | 9,489,953 | M1 realistic dataset (1:10 pos:neg) |
|
| 66 |
+
| `negbiodb_m1_uniform_random.parquet` | 467.3 MB | 1,767,380 | Control: uniform random negatives |
|
| 67 |
+
| `negbiodb_m1_degree_matched.parquet` | 274.9 MB | 1,767,380 | Control: degree-matched negatives |
|
| 68 |
+
| `negbiodb_m1_balanced_ddb.parquet` | 1.0 GB | 1,725,446 | M1 balanced with degree-balanced split |
|
| 69 |
+
|
| 70 |
+
### CT Files
|
| 71 |
+
| File | Size | Rows | Description |
|
| 72 |
+
|------|------|------|-------------|
|
| 73 |
+
| `ct/negbiodb_ct_pairs.parquet` | 1.7 MB | 102,850 | All CT failure pairs |
|
| 74 |
+
| `ct/negbiodb_ct_m1_balanced.parquet` | 308.7 KB | 11,222 | CT-M1 balanced (success vs failure) |
|
| 75 |
+
| `ct/negbiodb_ct_m1_realistic.parquet` | 896.4 KB | 36,957 | CT-M1 realistic ratio |
|
| 76 |
+
| `ct/negbiodb_ct_m2.parquet` | 2.2 MB | 112,298 | CT-M2 multiclass failure category |
|
| 77 |
+
|
| 78 |
+
### PPI Files
|
| 79 |
+
| File | Size | Rows | Description |
|
| 80 |
+
|------|------|------|-------------|
|
| 81 |
+
| `ppi/negbiodb_ppi_pairs.parquet` | 685.2 MB | 2,220,786 | All negative PPI pairs |
|
| 82 |
+
| `ppi/ppi_m1_balanced.parquet` | 120.5 MB | 123,456 | PPI-M1 balanced (1:1 pos:neg) |
|
| 83 |
+
| `ppi/ppi_m1_realistic.parquet` | 744.8 MB | 679,008 | PPI-M1 realistic (1:10 ratio) |
|
| 84 |
+
|
| 85 |
+
### LLM Benchmark Files (DTI)
|
| 86 |
+
| File | Items | Description |
|
| 87 |
+
|------|-------|-------------|
|
| 88 |
+
| `llm_benchmarks/l1_mcq.jsonl` | 1,942 | L1: 4-class activity MCQ |
|
| 89 |
+
| `llm_benchmarks/l3_reasoning_pilot.jsonl` | 50 | L3: Scientific reasoning (pilot) |
|
| 90 |
+
| `llm_benchmarks/l4_tested_untested.jsonl` | 500 | L4: Tested vs untested discrimination |
|
| 91 |
+
|
| 92 |
+
## Benchmark Tasks
|
| 93 |
+
|
| 94 |
+
### ML Benchmarks
|
| 95 |
+
| Task | Domain | Type | Splits |
|
| 96 |
+
|------|--------|------|--------|
|
| 97 |
+
| M1 | DTI | Binary (active/inactive) | random, cold_compound, cold_target, degree_balanced |
|
| 98 |
+
| CT-M1 | CT | Binary (success/failure) | random, cold_drug, cold_condition, temporal, scaffold, cold_both |
|
| 99 |
+
| CT-M2 | CT | 7-way failure category | Same as CT-M1 |
|
| 100 |
+
| PPI-M1 | PPI | Binary (interact/non-interact) | random, cold_protein, cold_both, degree_balanced |
|
| 101 |
+
|
| 102 |
+
### LLM Benchmarks (DTI)
|
| 103 |
+
| Task | Type | Size | Description |
|
| 104 |
+
|------|------|------|-------------|
|
| 105 |
+
| L1 | MCQ classification | 1,600 | 4-class activity level |
|
| 106 |
+
| L2 | Structured extraction | ~100 | Extract results from abstracts |
|
| 107 |
+
| L3 | Reasoning | 50 | Explain compound-target inactivity |
|
| 108 |
+
| L4 | Discrimination | 400 | Tested vs untested pair |
|
| 109 |
+
|
| 110 |
+
## License
|
| 111 |
+
|
| 112 |
+
This dataset is released under **CC BY-SA 4.0**, due to the viral clause of ChEMBL's CC BY-SA 3.0 license. See the LICENSE file for details.
|
| 113 |
+
|
| 114 |
+
## Citation
|
| 115 |
+
|
| 116 |
+
```bibtex
|
| 117 |
+
@dataset{negbiodb2026,
|
| 118 |
+
title={NegBioDB: A Negative Results Database for Biomedical Sciences},
|
| 119 |
+
author={Jang, Jungwon},
|
| 120 |
+
year={2026},
|
| 121 |
+
url={https://github.com/jang1563/NegBioDB}
|
| 122 |
+
}
|
| 123 |
+
```
|
data/compound_names.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d63e26b61ca7d3694ae20c9553d8460c827170557d114253d0e89b71fd5636f9
|
| 3 |
+
size 12400351
|
data/croissant.json
ADDED
|
@@ -0,0 +1,176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"@context": {
|
| 3 |
+
"@vocab": "https://schema.org/",
|
| 4 |
+
"sc": "https://schema.org/",
|
| 5 |
+
"cr": "http://mlcommons.org/croissant/",
|
| 6 |
+
"rai": "http://mlcommons.org/croissant/RAI/"
|
| 7 |
+
},
|
| 8 |
+
"@type": "sc:Dataset",
|
| 9 |
+
"name": "NegBioDB",
|
| 10 |
+
"description": "A large-scale database of experimentally confirmed negative results across three biomedical domains (DTI, Clinical Trials, PPI), with dual ML/LLM benchmarks.",
|
| 11 |
+
"license": "https://creativecommons.org/licenses/by-sa/4.0/",
|
| 12 |
+
"url": "https://github.com/jang1563/NegBioDB",
|
| 13 |
+
"version": "1.0.0",
|
| 14 |
+
"datePublished": "2026",
|
| 15 |
+
"creator": {
|
| 16 |
+
"@type": "sc:Person",
|
| 17 |
+
"name": "Jungwon Jang"
|
| 18 |
+
},
|
| 19 |
+
"distribution": [
|
| 20 |
+
{
|
| 21 |
+
"@type": "cr:FileObject",
|
| 22 |
+
"name": "dti_pairs",
|
| 23 |
+
"contentUrl": "negbiodb_dti_pairs.parquet",
|
| 24 |
+
"encodingFormat": "application/x-parquet",
|
| 25 |
+
"description": "All negative DTI pairs with source, tier, and activity data"
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"@type": "cr:FileObject",
|
| 29 |
+
"name": "dti_m1_balanced",
|
| 30 |
+
"contentUrl": "negbiodb_m1_balanced.parquet",
|
| 31 |
+
"encodingFormat": "application/x-parquet",
|
| 32 |
+
"description": "DTI M1 balanced benchmark dataset (1:1 positive:negative)"
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"@type": "cr:FileObject",
|
| 36 |
+
"name": "ct_pairs",
|
| 37 |
+
"contentUrl": "ct/negbiodb_ct_pairs.parquet",
|
| 38 |
+
"encodingFormat": "application/x-parquet",
|
| 39 |
+
"description": "All clinical trial failure pairs"
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"@type": "cr:FileObject",
|
| 43 |
+
"name": "ppi_pairs",
|
| 44 |
+
"contentUrl": "ppi/negbiodb_ppi_pairs.parquet",
|
| 45 |
+
"encodingFormat": "application/x-parquet",
|
| 46 |
+
"description": "All negative PPI pairs"
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"@type": "cr:FileObject",
|
| 50 |
+
"name": "llm_l1",
|
| 51 |
+
"contentUrl": "llm_benchmarks/l1_mcq.jsonl",
|
| 52 |
+
"encodingFormat": "application/jsonl",
|
| 53 |
+
"description": "L1 MCQ classification benchmark"
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"@type": "cr:FileObject",
|
| 57 |
+
"name": "llm_l4",
|
| 58 |
+
"contentUrl": "llm_benchmarks/l4_tested_untested.jsonl",
|
| 59 |
+
"encodingFormat": "application/jsonl",
|
| 60 |
+
"description": "L4 tested/untested discrimination benchmark"
|
| 61 |
+
}
|
| 62 |
+
],
|
| 63 |
+
"recordSet": [
|
| 64 |
+
{
|
| 65 |
+
"@type": "cr:RecordSet",
|
| 66 |
+
"name": "dti_pairs_record",
|
| 67 |
+
"source": "dti_pairs",
|
| 68 |
+
"field": [
|
| 69 |
+
{
|
| 70 |
+
"@type": "cr:Field",
|
| 71 |
+
"name": "inchikey_connectivity",
|
| 72 |
+
"dataType": "sc:Text",
|
| 73 |
+
"description": "inchikey connectivity"
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"@type": "cr:Field",
|
| 77 |
+
"name": "uniprot_id",
|
| 78 |
+
"dataType": "sc:Text",
|
| 79 |
+
"description": "uniprot id"
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"@type": "cr:Field",
|
| 83 |
+
"name": "activity_type",
|
| 84 |
+
"dataType": "sc:Text",
|
| 85 |
+
"description": "activity type"
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"@type": "cr:Field",
|
| 89 |
+
"name": "pchembl_value",
|
| 90 |
+
"dataType": "sc:Float",
|
| 91 |
+
"description": "pchembl value"
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"@type": "cr:Field",
|
| 95 |
+
"name": "source",
|
| 96 |
+
"dataType": "sc:Text",
|
| 97 |
+
"description": "source"
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"@type": "cr:Field",
|
| 101 |
+
"name": "tier",
|
| 102 |
+
"dataType": "sc:Text",
|
| 103 |
+
"description": "tier"
|
| 104 |
+
}
|
| 105 |
+
]
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"@type": "cr:RecordSet",
|
| 109 |
+
"name": "ct_pairs_record",
|
| 110 |
+
"source": "ct_pairs",
|
| 111 |
+
"field": [
|
| 112 |
+
{
|
| 113 |
+
"@type": "cr:Field",
|
| 114 |
+
"name": "nct_id",
|
| 115 |
+
"dataType": "sc:Text",
|
| 116 |
+
"description": "nct id"
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"@type": "cr:Field",
|
| 120 |
+
"name": "intervention_name",
|
| 121 |
+
"dataType": "sc:Text",
|
| 122 |
+
"description": "intervention name"
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"@type": "cr:Field",
|
| 126 |
+
"name": "failure_category",
|
| 127 |
+
"dataType": "sc:Text",
|
| 128 |
+
"description": "failure category"
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"@type": "cr:Field",
|
| 132 |
+
"name": "tier",
|
| 133 |
+
"dataType": "sc:Text",
|
| 134 |
+
"description": "tier"
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"@type": "cr:Field",
|
| 138 |
+
"name": "highest_phase_reached",
|
| 139 |
+
"dataType": "sc:Text",
|
| 140 |
+
"description": "highest phase reached"
|
| 141 |
+
}
|
| 142 |
+
]
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"@type": "cr:RecordSet",
|
| 146 |
+
"name": "ppi_pairs_record",
|
| 147 |
+
"source": "ppi_pairs",
|
| 148 |
+
"field": [
|
| 149 |
+
{
|
| 150 |
+
"@type": "cr:Field",
|
| 151 |
+
"name": "protein_a",
|
| 152 |
+
"dataType": "sc:Text",
|
| 153 |
+
"description": "protein a"
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"@type": "cr:Field",
|
| 157 |
+
"name": "protein_b",
|
| 158 |
+
"dataType": "sc:Text",
|
| 159 |
+
"description": "protein b"
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"@type": "cr:Field",
|
| 163 |
+
"name": "source",
|
| 164 |
+
"dataType": "sc:Text",
|
| 165 |
+
"description": "source"
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"@type": "cr:Field",
|
| 169 |
+
"name": "tier",
|
| 170 |
+
"dataType": "sc:Text",
|
| 171 |
+
"description": "tier"
|
| 172 |
+
}
|
| 173 |
+
]
|
| 174 |
+
}
|
| 175 |
+
]
|
| 176 |
+
}
|
data/ct/negbiodb_ct_m1_balanced.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d948f6ce736bd40ba8d0a65ae741ec3220aa26e29654d49487c0ffb257a97c39
|
| 3 |
+
size 316145
|
data/ct/negbiodb_ct_m1_degree_matched.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62d4e4469638dc4c7c45fd5e8992a3b8a8c45efe407b5790396622904ca827b0
|
| 3 |
+
size 428340
|
data/ct/negbiodb_ct_m1_realistic.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8fb6e691ec5dd69ad8ccbc4447f34981350a8bfb4f811c361b93e9d8cb0bcf7c
|
| 3 |
+
size 917905
|
data/ct/negbiodb_ct_m1_smiles_only.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95032b1277b1bb3f8740a568b4e4074e1e91b0aca815372721561d83a4ba1ece
|
| 3 |
+
size 174375
|
data/ct/negbiodb_ct_m1_uniform_random.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5136d02939f49b3ad2725a33924d4b83645c7aefdc68837983bd52dec4bcac3b
|
| 3 |
+
size 483222
|
data/ct/negbiodb_ct_m2.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7035bf7cf4aecd8b224e7bd8f888dd1c8fd4e5b9ce8d44c7aacb59d189e2a7aa
|
| 3 |
+
size 2313878
|
data/ct/negbiodb_ct_pairs.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3334c220fa96232e83437683eec7e441e5fca3b313161883aa363217bf99eb88
|
| 3 |
+
size 1752110
|
data/ct_llm/ct_l1_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ct_llm/ct_l1_metadata.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ct-l1",
|
| 3 |
+
"created": "2026-03-18T18:44:59.269332+00:00",
|
| 4 |
+
"total": 1500,
|
| 5 |
+
"classes": {
|
| 6 |
+
"A": 300,
|
| 7 |
+
"B": 300,
|
| 8 |
+
"C": 300,
|
| 9 |
+
"D": 300,
|
| 10 |
+
"E": 300
|
| 11 |
+
},
|
| 12 |
+
"difficulty": {
|
| 13 |
+
"hard": 856,
|
| 14 |
+
"medium": 524,
|
| 15 |
+
"easy": 120
|
| 16 |
+
},
|
| 17 |
+
"splits": {
|
| 18 |
+
"fewshot": 300,
|
| 19 |
+
"val": 300,
|
| 20 |
+
"test": 900
|
| 21 |
+
}
|
| 22 |
+
}
|
data/ct_llm/ct_l2_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ct_llm/ct_l2_metadata.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ct-l2",
|
| 3 |
+
"created": "2026-03-18T18:43:54.190316+00:00",
|
| 4 |
+
"total": 500,
|
| 5 |
+
"categories": {
|
| 6 |
+
"other": 69,
|
| 7 |
+
"enrollment": 185,
|
| 8 |
+
"safety": 36,
|
| 9 |
+
"efficacy": 90,
|
| 10 |
+
"regulatory": 10,
|
| 11 |
+
"strategic": 84,
|
| 12 |
+
"design": 26
|
| 13 |
+
},
|
| 14 |
+
"difficulty": {
|
| 15 |
+
"hard": 35,
|
| 16 |
+
"easy": 115,
|
| 17 |
+
"medium": 350
|
| 18 |
+
},
|
| 19 |
+
"splits": {
|
| 20 |
+
"fewshot": 50,
|
| 21 |
+
"val": 50,
|
| 22 |
+
"test": 400
|
| 23 |
+
}
|
| 24 |
+
}
|
data/ct_llm/ct_l3_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ct_llm/ct_l3_metadata.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ct-l3",
|
| 3 |
+
"created": "2026-03-18T18:43:01.086284+00:00",
|
| 4 |
+
"total": 200,
|
| 5 |
+
"categories": {
|
| 6 |
+
"efficacy": 200
|
| 7 |
+
},
|
| 8 |
+
"splits": {
|
| 9 |
+
"fewshot": 20,
|
| 10 |
+
"val": 20,
|
| 11 |
+
"test": 160
|
| 12 |
+
}
|
| 13 |
+
}
|
data/ct_llm/ct_l4_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ct_llm/ct_l4_metadata.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ct-l4",
|
| 3 |
+
"created": "2026-03-18T18:46:58.238502+00:00",
|
| 4 |
+
"total": 500,
|
| 5 |
+
"classes": {
|
| 6 |
+
"tested": 250,
|
| 7 |
+
"untested": 250
|
| 8 |
+
},
|
| 9 |
+
"temporal": {
|
| 10 |
+
"post_2023": 125,
|
| 11 |
+
"pre_2020": 125
|
| 12 |
+
},
|
| 13 |
+
"splits": {
|
| 14 |
+
"fewshot": 50,
|
| 15 |
+
"val": 50,
|
| 16 |
+
"test": 400
|
| 17 |
+
},
|
| 18 |
+
"n_leaks": 0
|
| 19 |
+
}
|
data/ge_llm/ge-l1_metadata.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ge-l1",
|
| 3 |
+
"domain": "ge",
|
| 4 |
+
"n_total": 1200,
|
| 5 |
+
"n_per_class": {
|
| 6 |
+
"common_essential": 300,
|
| 7 |
+
"selective_essential": 300,
|
| 8 |
+
"non_essential": 300,
|
| 9 |
+
"unknown_untested": 300
|
| 10 |
+
},
|
| 11 |
+
"difficulty_distribution": {
|
| 12 |
+
"easy": "480",
|
| 13 |
+
"medium": "420",
|
| 14 |
+
"hard": "300"
|
| 15 |
+
},
|
| 16 |
+
"split_distribution": {
|
| 17 |
+
"test": "720",
|
| 18 |
+
"fewshot": "240",
|
| 19 |
+
"val": "240"
|
| 20 |
+
},
|
| 21 |
+
"seed": 42
|
| 22 |
+
}
|
data/ge_llm/ge-l2_metadata.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ge-l2",
|
| 3 |
+
"domain": "ge",
|
| 4 |
+
"n_total": 500,
|
| 5 |
+
"design": "constructed_evidence",
|
| 6 |
+
"gene_distribution": {
|
| 7 |
+
"1": "200",
|
| 8 |
+
"2": "200",
|
| 9 |
+
"3": "100"
|
| 10 |
+
},
|
| 11 |
+
"split_distribution": {
|
| 12 |
+
"test": "400",
|
| 13 |
+
"fewshot": "50",
|
| 14 |
+
"val": "50"
|
| 15 |
+
},
|
| 16 |
+
"seed": 42
|
| 17 |
+
}
|
data/ge_llm/ge-l3_metadata.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ge-l3",
|
| 3 |
+
"domain": "ge",
|
| 4 |
+
"n_total": 200,
|
| 5 |
+
"ne_type_distribution": {
|
| 6 |
+
"reference": 100,
|
| 7 |
+
"context": 100
|
| 8 |
+
},
|
| 9 |
+
"split_distribution": {
|
| 10 |
+
"test": "160",
|
| 11 |
+
"val": "20",
|
| 12 |
+
"fewshot": "20"
|
| 13 |
+
},
|
| 14 |
+
"seed": 42
|
| 15 |
+
}
|
data/ge_llm/ge-l4_metadata.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ge-l4",
|
| 3 |
+
"domain": "ge",
|
| 4 |
+
"n_total": 475,
|
| 5 |
+
"n_tested": 250,
|
| 6 |
+
"n_untested": 225,
|
| 7 |
+
"n_tested_old": 125,
|
| 8 |
+
"n_tested_new": 125,
|
| 9 |
+
"n_untested_trick": 125,
|
| 10 |
+
"n_untested_obvious": 100,
|
| 11 |
+
"split_distribution": {
|
| 12 |
+
"fewshot": 50,
|
| 13 |
+
"val": 50,
|
| 14 |
+
"test": 375
|
| 15 |
+
},
|
| 16 |
+
"seed": 42
|
| 17 |
+
}
|
data/ge_llm/ge_l1_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ge_llm/ge_l2_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ge_llm/ge_l3_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ge_llm/ge_l4_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/llm_benchmarks/l1_mcq.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/llm_benchmarks/l2_candidates.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/llm_benchmarks/l3_reasoning_pilot.jsonl
ADDED
|
@@ -0,0 +1,50 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"class": "reasoning", "compound_name": "(S)-N-(1-((5-(N-(tert- butyl)sulfamoyl)naphthalen-1- yl)amino)-1-oxo-3-phenylpropan- 2-yl)pyrimidine-2-carboxamide::US10968172, ID # 4o", "compound_smiles": "CC(C)(C)NS(=O)(=O)c1cccc2c(NC(=O)[C@H](Cc3ccccc3)NC(=O)c3ncccn3)cccc12", "compound_inchikey": "APIMXUDTMIUDED-QHCPKHFHSA-N", "target_uniprot": "Q70CQ3", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: (S)-N-(1-((5-(N-(tert- butyl)sulfamoyl)naphthalen-1- yl)amino)-1-oxo-3-phenylpropan- 2-yl)pyrimidine-2-carboxamide::US10968172, ID # 4o\nSMILES: CC(C)(C)NS(=O)(=O)c1cccc2c(NC(=O)[C@H](Cc3ccccc3)NC(=O)c3ncccn3)cccc12\nTarget: Q70CQ3, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "fewshot", "question_id": "L3-0000"}
|
| 2 |
+
{"class": "reasoning", "compound_name": "US9409892, 33", "compound_smiles": "N#Cc1nc(C(=O)NC2CCC[C@H]2C(=O)O)c(O)c2ccc(Oc3ccccc3)cc12", "compound_inchikey": "TYSUUCOXMBQYMY-PYUWXLGESA-N", "target_uniprot": "Q96KS0", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: US9409892, 33\nSMILES: N#Cc1nc(C(=O)NC2CCC[C@H]2C(=O)O)c(O)c2ccc(Oc3ccccc3)cc12\nTarget: Q96KS0, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "fewshot", "question_id": "L3-0001"}
|
| 3 |
+
{"class": "reasoning", "compound_name": "US9434724, 24", "compound_smiles": "O=C(OCC12CCN(CC1)CC2)c1c[nH]c2cccc(Br)c12", "compound_inchikey": "MRUNYBDRMJNCIJ-UHFFFAOYSA-N", "target_uniprot": "P11712", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: US9434724, 24\nSMILES: O=C(OCC12CCN(CC1)CC2)c1c[nH]c2cccc(Br)c12\nTarget: P11712, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "fewshot", "question_id": "L3-0002"}
|
| 4 |
+
{"class": "reasoning", "compound_name": "3-methyl-6-(piperidin-1-yl)-2H-naphtho[1,2,3-de]quinoline-2,7(3H)-dione", "compound_smiles": "Cn1c(=O)cc2c3c(c(N4CCCCC4)ccc31)C(=O)c1ccccc1-2", "compound_inchikey": "ZMXVWFFEPNMQNW-UHFFFAOYSA-N", "target_uniprot": "P02675", "target_gene": null, "target_family": "protein", "num_assays": 3, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: 3-methyl-6-(piperidin-1-yl)-2H-naphtho[1,2,3-de]quinoline-2,7(3H)-dione\nSMILES: Cn1c(=O)cc2c3c(c(N4CCCCC4)ccc31)C(=O)c1ccccc1-2\nTarget: P02675, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "fewshot", "question_id": "L3-0003"}
|
| 5 |
+
{"class": "reasoning", "compound_name": "US9120812, 155", "compound_smiles": "CCN1CC(C)(C)Oc2nc(N3CC4CCC(C3)O4)nc(-c3ccc(NC(=O)Nc4cc(OC)c(F)cc4F)cc3)c2C1=O", "compound_inchikey": "GWMPVSYXOSFCEQ-UHFFFAOYSA-N", "target_uniprot": "P27986", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: US9120812, 155\nSMILES: CCN1CC(C)(C)Oc2nc(N3CC4CCC(C3)O4)nc(-c3ccc(NC(=O)Nc4cc(OC)c(F)cc4F)cc3)c2C1=O\nTarget: P27986, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "fewshot", "question_id": "L3-0004"}
|
| 6 |
+
{"class": "reasoning", "compound_name": "DINACICLIB", "compound_smiles": "CCc1cnn2c(NCc3ccc[n+]([O-])c3)cc(N3CCCC[C@H]3CCO)nc12", "compound_inchikey": "PIMQWRZWLQKKBJ-SFHVURJKSA-N", "target_uniprot": "P36507", "target_gene": "MEK2", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: DINACICLIB\nSMILES: CCc1cnn2c(NCc3ccc[n+]([O-])c3)cc(N3CCCC[C@H]3CCO)nc12\nTarget: MEK2 (P36507), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "val", "question_id": "L3-0005"}
|
| 7 |
+
{"class": "reasoning", "compound_name": "US9149492, 16", "compound_smiles": "CCCCCCCCCCNc1cc(-c2ccccc2)nn1-c1ccccc1", "compound_inchikey": "CJKKEVCKJUHCKP-UHFFFAOYSA-N", "target_uniprot": "O75908", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: US9149492, 16\nSMILES: CCCCCCCCCCNc1cc(-c2ccccc2)nn1-c1ccccc1\nTarget: O75908, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "val", "question_id": "L3-0006"}
|
| 8 |
+
{"class": "reasoning", "compound_name": "3-phenyl-2H-benzo[4,5]thiazolo[3,2-a][1,3,5]triazine-2,4(3H)-dione", "compound_smiles": "O=c1nc2sc3ccccc3n2c(=O)n1-c1ccccc1", "compound_inchikey": "FCPXRIBIDATENX-UHFFFAOYSA-N", "target_uniprot": "O76083", "target_gene": null, "target_family": "protein", "num_assays": 3, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: 3-phenyl-2H-benzo[4,5]thiazolo[3,2-a][1,3,5]triazine-2,4(3H)-dione\nSMILES: O=c1nc2sc3ccccc3n2c(=O)n1-c1ccccc1\nTarget: O76083, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "val", "question_id": "L3-0007"}
|
| 9 |
+
{"class": "reasoning", "compound_name": "HYDROXYFASUDIL", "compound_smiles": "O=S(=O)(c1cccc2c(O)nccc12)N1CCCNCC1", "compound_inchikey": "ZAVGJDAFCZAWSZ-UHFFFAOYSA-N", "target_uniprot": "Q14683", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: HYDROXYFASUDIL\nSMILES: O=S(=O)(c1cccc2c(O)nccc12)N1CCCNCC1\nTarget: Q14683, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "val", "question_id": "L3-0008"}
|
| 10 |
+
{"class": "reasoning", "compound_name": "TERFENADINE", "compound_smiles": "CC(C)(C)c1ccc(C(O)CCCN2CCC(C(O)(c3ccccc3)c3ccccc3)CC2)cc1", "compound_inchikey": "GUGOEEXESWIERI-UHFFFAOYSA-N", "target_uniprot": "Q03431", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 1, "evidence_quality": "silver", "context_text": "Compound: TERFENADINE\nSMILES: CC(C)(C)c1ccc(C(O)CCCN2CCC(C(O)(c3ccccc3)c3ccccc3)CC2)cc1\nTarget: Q03431, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "val", "question_id": "L3-0009"}
|
| 11 |
+
{"class": "reasoning", "compound_name": "ZARDAVERINE", "compound_smiles": "COc1cc(-c2ccc(=O)[nH]n2)ccc1OC(F)F", "compound_inchikey": "HJMQDJPMQIHLPB-UHFFFAOYSA-N", "target_uniprot": "Q9HCR9", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: ZARDAVERINE\nSMILES: COc1cc(-c2ccc(=O)[nH]n2)ccc1OC(F)F\nTarget: Q9HCR9, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0010"}
|
| 12 |
+
{"class": "reasoning", "compound_name": "3-chloro-N-{2-[6-(trifluoromethyl)pyridin-2-yl]-7H-pyrrolo[2,3-d]pyrimidin-4-yl}pyridin-4-amine::US10399987, Example 40", "compound_smiles": "FC(F)(F)c1cccc(-c2nc(Nc3ccncc3Cl)c3cc[nH]c3n2)n1", "compound_inchikey": "WDJWKQYJNPIICW-UHFFFAOYSA-N", "target_uniprot": "P37173", "target_gene": "TGFBR2", "target_family": "kinase", "num_assays": 2, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: 3-chloro-N-{2-[6-(trifluoromethyl)pyridin-2-yl]-7H-pyrrolo[2,3-d]pyrimidin-4-yl}pyridin-4-amine::US10399987, Example 40\nSMILES: FC(F)(F)c1cccc(-c2nc(Nc3ccncc3Cl)c3cc[nH]c3n2)n1\nTarget: TGFBR2 (P37173), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0011"}
|
| 13 |
+
{"class": "reasoning", "compound_name": "4-(4-Methyl-pyridin-2-ylamino)-piperidine-1-carboxylic acid ethyl ester", "compound_smiles": "CCOC(=O)N1CCC(Nc2cc(C)ccn2)CC1", "compound_inchikey": "LNRMJBWADUSJTA-UHFFFAOYSA-N", "target_uniprot": "P29475", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: 4-(4-Methyl-pyridin-2-ylamino)-piperidine-1-carboxylic acid ethyl ester\nSMILES: CCOC(=O)N1CCC(Nc2cc(C)ccn2)CC1\nTarget: P29475, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0012"}
|
| 14 |
+
{"class": "reasoning", "compound_name": "PF-04217903", "compound_smiles": "OCCn1cc(-c2cnc3nnn(Cc4ccc5ncccc5c4)c3n2)cn1", "compound_inchikey": "PDMUGYOXRHVNMO-UHFFFAOYSA-N", "target_uniprot": "O15197", "target_gene": "EPHB6", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: PF-04217903\nSMILES: OCCn1cc(-c2cnc3nnn(Cc4ccc5ncccc5c4)c3n2)cn1\nTarget: EPHB6 (O15197), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0013"}
|
| 15 |
+
{"class": "reasoning", "compound_name": "BARICITINIB", "compound_smiles": "CCS(=O)(=O)N1CC(CC#N)(n2cc(-c3ncnc4[nH]ccc34)cn2)C1", "compound_inchikey": "XUZMWHLSFXCVMG-UHFFFAOYSA-N", "target_uniprot": "P00519", "target_gene": "ABL1", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: BARICITINIB\nSMILES: CCS(=O)(=O)N1CC(CC#N)(n2cc(-c3ncnc4[nH]ccc34)cn2)C1\nTarget: ABL1 (P00519), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0014"}
|
| 16 |
+
{"class": "reasoning", "compound_name": "US9163008, 71", "compound_smiles": "C[C@H](Nc1cc(C(N)=O)nc(N2CCN(C(c3ccc(F)cc3)c3ccc(F)cc3)CC2)n1)C(N)=O", "compound_inchikey": "SSLDCNYHPUYQRB-HNNXBMFYSA-N", "target_uniprot": "Q15858", "target_gene": null, "target_family": "protein", "num_assays": 3, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: US9163008, 71\nSMILES: C[C@H](Nc1cc(C(N)=O)nc(N2CCN(C(c3ccc(F)cc3)c3ccc(F)cc3)CC2)n1)C(N)=O\nTarget: Q15858, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0015"}
|
| 17 |
+
{"class": "reasoning", "compound_name": "BAFETINIB", "compound_smiles": "Cc1ccc(NC(=O)c2ccc(CN3CC[C@H](N(C)C)C3)c(C(F)(F)F)c2)cc1Nc1nccc(-c2cncnc2)n1", "compound_inchikey": "ZGBAJMQHJDFTQJ-DEOSSOPVSA-N", "target_uniprot": "Q14839", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: BAFETINIB\nSMILES: Cc1ccc(NC(=O)c2ccc(CN3CC[C@H](N(C)C)C3)c(C(F)(F)F)c2)cc1Nc1nccc(-c2cncnc2)n1\nTarget: Q14839, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0016"}
|
| 18 |
+
{"class": "reasoning", "compound_name": "LENVATINIB", "compound_smiles": "COc1cc2nccc(Oc3ccc(NC(=O)NC4CC4)c(Cl)c3)c2cc1C(N)=O", "compound_inchikey": "WOSKHXYHFSIKNG-UHFFFAOYSA-N", "target_uniprot": "P31751", "target_gene": "AKT2", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: LENVATINIB\nSMILES: COc1cc2nccc(Oc3ccc(NC(=O)NC4CC4)c(Cl)c3)c2cc1C(N)=O\nTarget: AKT2 (P31751), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0017"}
|
| 19 |
+
{"class": "reasoning", "compound_name": "2-((4-(5-chloropyrimidin-2-yl)piperazin-1-yl)methyl)-1-methyl-1H-benzo[d]imidazole", "compound_smiles": "Cn1c(CN2CCN(c3ncc(Cl)cn3)CC2)nc2ccccc21", "compound_inchikey": "IXXZGBOSQDVMTC-UHFFFAOYSA-N", "target_uniprot": "Q13255", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 1, "evidence_quality": "silver", "context_text": "Compound: 2-((4-(5-chloropyrimidin-2-yl)piperazin-1-yl)methyl)-1-methyl-1H-benzo[d]imidazole\nSMILES: Cn1c(CN2CCN(c3ncc(Cl)cn3)CC2)nc2ccccc21\nTarget: Q13255, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0018"}
|
| 20 |
+
{"class": "reasoning", "compound_name": "SID17413282", "compound_smiles": "Cc1ccc(NNS(=O)(=O)c2ccc(C)cc2)cc1", "compound_inchikey": "OSDZCCMUWCNEOP-UHFFFAOYSA-N", "target_uniprot": "O14786", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: SID17413282\nSMILES: Cc1ccc(NNS(=O)(=O)c2ccc(C)cc2)cc1\nTarget: O14786, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0019"}
|
| 21 |
+
{"class": "reasoning", "compound_name": "SID853690", "compound_smiles": "CCCCNC(=O)C1CCCN1C(=O)Nc1cccc(OC)c1", "compound_inchikey": "QUPOBSOOZUZGRS-UHFFFAOYSA-N", "target_uniprot": "P04062", "target_gene": null, "target_family": "protein", "num_assays": 3, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: SID853690\nSMILES: CCCCNC(=O)C1CCCN1C(=O)Nc1cccc(OC)c1\nTarget: P04062, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0020"}
|
| 22 |
+
{"class": "reasoning", "compound_name": "pyrrolidin-1-yl-[(2S,4S)-4-(5-trifluoromethyl-1,3-dihydro-isoindole-2-carbonyl)-pyrrolidin-2-yl]-methanone", "compound_smiles": "O=C([C@@H]1CN[C@H](C(=O)N2CCCC2)C1)N1Cc2ccc(C(F)(F)F)cc2C1", "compound_inchikey": "ZXYMQVABWZXTNI-BBRMVZONSA-N", "target_uniprot": "Q6V1X1", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: pyrrolidin-1-yl-[(2S,4S)-4-(5-trifluoromethyl-1,3-dihydro-isoindole-2-carbonyl)-pyrrolidin-2-yl]-methanone\nSMILES: O=C([C@@H]1CN[C@H](C(=O)N2CCCC2)C1)N1Cc2ccc(C(F)(F)F)cc2C1\nTarget: Q6V1X1, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0021"}
|
| 23 |
+
{"class": "reasoning", "compound_name": "OSIMERTINIB", "compound_smiles": "C=CC(=O)Nc1cc(Nc2nccc(-c3cn(C)c4ccccc34)n2)c(OC)cc1N(C)CCN(C)C", "compound_inchikey": "DUYJMQONPNNFPI-UHFFFAOYSA-N", "target_uniprot": "O14578", "target_gene": "CIT", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: OSIMERTINIB\nSMILES: C=CC(=O)Nc1cc(Nc2nccc(-c3cn(C)c4ccccc34)n2)c(OC)cc1N(C)CCN(C)C\nTarget: CIT (O14578), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0022"}
|
| 24 |
+
{"class": "reasoning", "compound_name": "CEP-32496", "compound_smiles": "COc1cc2ncnc(Oc3cccc(NC(=O)Nc4cc(C(C)(C)C(F)(F)F)on4)c3)c2cc1OC", "compound_inchikey": "DKNUPRMJNUQNHR-UHFFFAOYSA-N", "target_uniprot": "P36896", "target_gene": "ACVR1B", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: CEP-32496\nSMILES: COc1cc2ncnc(Oc3cccc(NC(=O)Nc4cc(C(C)(C)C(F)(F)F)on4)c3)c2cc1OC\nTarget: ACVR1B (P36896), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0023"}
|
| 25 |
+
{"class": "reasoning", "compound_name": "SID7968677", "compound_smiles": "Cc1cc(NC(=O)CSc2ncnc3ccccc23)no1", "compound_inchikey": "KDDOCAGTMUZHJD-UHFFFAOYSA-N", "target_uniprot": "P11142", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: SID7968677\nSMILES: Cc1cc(NC(=O)CSc2ncnc3ccccc23)no1\nTarget: P11142, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0024"}
|
| 26 |
+
{"class": "reasoning", "compound_name": "PACRITINIB", "compound_smiles": "C1=N/C2=N/c3ccc(OCCN4CCCC4)c(c3)COC/C=C/COCc3cccc(c3)C(=C1)N2", "compound_inchikey": "HWXVIOGONBBTBY-ONEGZZNKSA-N", "target_uniprot": "O75460", "target_gene": "ERN1", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: PACRITINIB\nSMILES: C1=N/C2=N/c3ccc(OCCN4CCCC4)c(c3)COC/C=C/COCc3cccc(c3)C(=C1)N2\nTarget: ERN1 (O75460), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0025"}
|
| 27 |
+
{"class": "reasoning", "compound_name": "ACTB-1003", "compound_smiles": "COCc1c(-c2ccc(NC(=O)Nc3cc(C(F)(F)F)ccc3F)c(F)c2)c2c(N)ncnn2c1CN1CCOCC1", "compound_inchikey": "GZPJCJKUZPUFAL-UHFFFAOYSA-N", "target_uniprot": "P78368", "target_gene": "CSNK1G2", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: ACTB-1003\nSMILES: COCc1c(-c2ccc(NC(=O)Nc3cc(C(F)(F)F)ccc3F)c(F)c2)c2c(N)ncnn2c1CN1CCOCC1\nTarget: CSNK1G2 (P78368), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0026"}
|
| 28 |
+
{"class": "reasoning", "compound_name": "VEMURAFENIB", "compound_smiles": "CCCS(=O)(=O)Nc1ccc(F)c(C(=O)c2c[nH]c3ncc(-c4ccc(Cl)cc4)cc23)c1F", "compound_inchikey": "GPXBXXGIAQBQNI-UHFFFAOYSA-N", "target_uniprot": "Q5VT25", "target_gene": "MRCKA", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: VEMURAFENIB\nSMILES: CCCS(=O)(=O)Nc1ccc(F)c(C(=O)c2c[nH]c3ncc(-c4ccc(Cl)cc4)cc23)c1F\nTarget: MRCKA (Q5VT25), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0027"}
|
| 29 |
+
{"class": "reasoning", "compound_name": "BMS-690514", "compound_smiles": "COc1cccc(Nc2ncnn3ccc(CN4CC[C@@H](N)[C@H](O)C4)c23)c1", "compound_inchikey": "CSGQVNMSRKWUSH-IAGOWNOFSA-N", "target_uniprot": "Q709F0", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: BMS-690514\nSMILES: COc1cccc(Nc2ncnn3ccc(CN4CC[C@@H](N)[C@H](O)C4)c23)c1\nTarget: Q709F0, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0028"}
|
| 30 |
+
{"class": "reasoning", "compound_name": "AZD-1480", "compound_smiles": "Cc1cc(Nc2nc(N[C@@H](C)c3ncc(F)cn3)ncc2Cl)[nH]n1", "compound_inchikey": "PDOQBOJDRPLBQU-QMMMGPOBSA-N", "target_uniprot": "Q9NSD9", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: AZD-1480\nSMILES: Cc1cc(Nc2nc(N[C@@H](C)c3ncc(F)cn3)ncc2Cl)[nH]n1\nTarget: Q9NSD9, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0029"}
|
| 31 |
+
{"class": "reasoning", "compound_name": "3-(5-(((1S,2S)-2-(ethylamino) cyclohexyl)amino)- 1-oxoisoindolin-2-yl)piperidine-2,6-dione::US11185537, Compound I-40", "compound_smiles": "CCN[C@H]1CCCC[C@@H]1Nc1ccc2c(c1)CN(C1CCC(=O)NC1=O)C2=O", "compound_inchikey": "DDATULQPTHGNEF-UBFHEZILSA-N", "target_uniprot": "Q13422", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: 3-(5-(((1S,2S)-2-(ethylamino) cyclohexyl)amino)- 1-oxoisoindolin-2-yl)piperidine-2,6-dione::US11185537, Compound I-40\nSMILES: CCN[C@H]1CCCC[C@@H]1Nc1ccc2c(c1)CN(C1CCC(=O)NC1=O)C2=O\nTarget: Q13422, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0030"}
|
| 32 |
+
{"class": "reasoning", "compound_name": "SID26658848", "compound_smiles": "C=CCOc1ccc(/C=C2\\SC(=S)N(CCC(=O)O)C2=O)cc1", "compound_inchikey": "AEZGVSCOIZTTBC-RAXLEYEMSA-N", "target_uniprot": "Q9Y6Q9", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: SID26658848\nSMILES: C=CCOc1ccc(/C=C2\\SC(=S)N(CCC(=O)O)C2=O)cc1\nTarget: Q9Y6Q9, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0031"}
|
| 33 |
+
{"class": "reasoning", "compound_name": "US9981950, Example 13", "compound_smiles": "Clc1ccc(-n2cc(Cn3cnc(C4CC4)n3)nn2)cc1", "compound_inchikey": "FYLLXYYSVQPAKZ-UHFFFAOYSA-N", "target_uniprot": "Q12809", "target_gene": null, "target_family": "protein", "num_assays": 3, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: US9981950, Example 13\nSMILES: Clc1ccc(-n2cc(Cn3cnc(C4CC4)n3)nn2)cc1\nTarget: Q12809, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0032"}
|
| 34 |
+
{"class": "reasoning", "compound_name": "APITOLISIB", "compound_smiles": "Cc1c(CN2CCN(C(=O)[C@H](C)O)CC2)sc2c(N3CCOCC3)nc(-c3cnc(N)nc3)nc12", "compound_inchikey": "YOVVNQKCSKSHKT-HNNXBMFYSA-N", "target_uniprot": "Q92918", "target_gene": "HPK1", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: APITOLISIB\nSMILES: Cc1c(CN2CCN(C(=O)[C@H](C)O)CC2)sc2c(N3CCOCC3)nc(-c3cnc(N)nc3)nc12\nTarget: HPK1 (Q92918), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0033"}
|
| 35 |
+
{"class": "reasoning", "compound_name": "THIRAM", "compound_smiles": "CN(C)C(=S)SSC(=S)N(C)C", "compound_inchikey": "KUAZQDVKQLNFPE-UHFFFAOYSA-N", "target_uniprot": "Q6W5P4", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 1, "evidence_quality": "silver", "context_text": "Compound: THIRAM\nSMILES: CN(C)C(=S)SSC(=S)N(C)C\nTarget: Q6W5P4, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0034"}
|
| 36 |
+
{"class": "reasoning", "compound_name": "SID22402056", "compound_smiles": "O=C(CCl)NC(=O)NC12CC3CC(CC(C3)C1)C2", "compound_inchikey": "OBLFEICGDWLKQF-UHFFFAOYSA-N", "target_uniprot": "O15492", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: SID22402056\nSMILES: O=C(CCl)NC(=O)NC12CC3CC(CC(C3)C1)C2\nTarget: O15492, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0035"}
|
| 37 |
+
{"class": "reasoning", "compound_name": "COBIMETINIB", "compound_smiles": "O=C(c1ccc(F)c(F)c1Nc1ccc(I)cc1F)N1CC(O)([C@@H]2CCCCN2)C1", "compound_inchikey": "BSMCAPRUBJMWDF-KRWDZBQOSA-N", "target_uniprot": "Q7Z406", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: COBIMETINIB\nSMILES: O=C(c1ccc(F)c(F)c1Nc1ccc(I)cc1F)N1CC(O)([C@@H]2CCCCN2)C1\nTarget: Q7Z406, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0036"}
|
| 38 |
+
{"class": "reasoning", "compound_name": "2-(3-chloro-N-(2-chloroacetyl)-4-methoxy-anilino)-N-(2-phenylethyl)-2-(2-thienyl)acetamide", "compound_smiles": "COc1ccc(N(C(=O)CCl)C(C(=O)NCCc2ccccc2)c2cccs2)cc1Cl", "compound_inchikey": "UNVKYJSNMVDZJE-UHFFFAOYSA-N", "target_uniprot": "P41145", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 1, "evidence_quality": "silver", "context_text": "Compound: 2-(3-chloro-N-(2-chloroacetyl)-4-methoxy-anilino)-N-(2-phenylethyl)-2-(2-thienyl)acetamide\nSMILES: COc1ccc(N(C(=O)CCl)C(C(=O)NCCc2ccccc2)c2cccs2)cc1Cl\nTarget: P41145, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0037"}
|
| 39 |
+
{"class": "reasoning", "compound_name": "US11584714, Compound 1471", "compound_smiles": "C[C@H](NC(=O)[C@H]1C[C@H](c2ccccc2)CN1)C(=O)NCc1cnc2sccc2c1", "compound_inchikey": "QOGGVKMWFMLWTF-ZOCIIQOWSA-N", "target_uniprot": "P00734", "target_gene": null, "target_family": "protein", "num_assays": 3, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: US11584714, Compound 1471\nSMILES: C[C@H](NC(=O)[C@H]1C[C@H](c2ccccc2)CN1)C(=O)NCc1cnc2sccc2c1\nTarget: P00734, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0038"}
|
| 40 |
+
{"class": "reasoning", "compound_name": "SID24780368", "compound_smiles": "Cc1ccc(C)c(-n2cc(CNCCc3cnccn3)c(-c3ccccc3)n2)c1", "compound_inchikey": "FFAQZBJIGLDKCL-UHFFFAOYSA-N", "target_uniprot": "P41143", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: SID24780368\nSMILES: Cc1ccc(C)c(-n2cc(CNCCc3cnccn3)c(-c3ccccc3)n2)c1\nTarget: P41143, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0039"}
|
| 41 |
+
{"class": "reasoning", "compound_name": "(R)-7-fluoro-6-methoxy-N-(1'-azaspiro[cyclopropane-1,2'-bicyclo[2.2.2]octan]-3'-yl)benzo[b]thiophene-2-carboxamide::US10183938, Compound (R)-155", "compound_smiles": "COc1ccc2cc(C(=O)N[C@@H]3C4CCN(CC4)C34CC4)sc2c1F", "compound_inchikey": "WIHCHZXZKLIHFS-QGZVFWFLSA-N", "target_uniprot": "P46098", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: (R)-7-fluoro-6-methoxy-N-(1'-azaspiro[cyclopropane-1,2'-bicyclo[2.2.2]octan]-3'-yl)benzo[b]thiophene-2-carboxamide::US10183938, Compound (R)-155\nSMILES: COc1ccc2cc(C(=O)N[C@@H]3C4CCN(CC4)C34CC4)sc2c1F\nTarget: P46098, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0040"}
|
| 42 |
+
{"class": "reasoning", "compound_name": "SID26665079", "compound_smiles": "COc1ccc2c3c([nH]c2c1)C1CC2C(CC(O)C(OC)C2CO)CN1CC3", "compound_inchikey": "YCLAQLTVHNAEEQ-UHFFFAOYSA-N", "target_uniprot": "Q9NZJ5", "target_gene": null, "target_family": "protein", "num_assays": 3, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: SID26665079\nSMILES: COc1ccc2c3c([nH]c2c1)C1CC2C(CC(O)C(OC)C2CO)CN1CC3\nTarget: Q9NZJ5, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0041"}
|
| 43 |
+
{"class": "reasoning", "compound_name": "MERESTINIB", "compound_smiles": "Cc1ccc(C(=O)Nc2ccc(Oc3cc4cnn(C)c4cc3-c3cn[nH]c3)c(F)c2)c(=O)n1-c1ccc(F)cc1", "compound_inchikey": "QHADVLVFMKEIIP-UHFFFAOYSA-N", "target_uniprot": "P23443", "target_gene": "S6K1", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: MERESTINIB\nSMILES: Cc1ccc(C(=O)Nc2ccc(Oc3cc4cnn(C)c4cc3-c3cn[nH]c3)c(F)c2)c(=O)n1-c1ccc(F)cc1\nTarget: S6K1 (P23443), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0042"}
|
| 44 |
+
{"class": "reasoning", "compound_name": "SID49645791", "compound_smiles": "Cc1coc2c1c(C)cc1oc(=O)c(CC(=O)NCCCCCC(=O)O)c(C)c12", "compound_inchikey": "CSLQCSIPGCWTGF-UHFFFAOYSA-N", "target_uniprot": "P49862", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 3, "evidence_quality": "silver", "context_text": "Compound: SID49645791\nSMILES: Cc1coc2c1c(C)cc1oc(=O)c(CC(=O)NCCCCCC(=O)O)c(C)c12\nTarget: P49862, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0043"}
|
| 45 |
+
{"class": "reasoning", "compound_name": "SID17412990", "compound_smiles": "CC(=O)NS(=O)(=O)c1ccc(NC(=S)NC(=O)c2ccco2)cc1", "compound_inchikey": "JBEDJYDIYZFRRT-UHFFFAOYSA-N", "target_uniprot": "O75496", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 1, "evidence_quality": "silver", "context_text": "Compound: SID17412990\nSMILES: CC(=O)NS(=O)(=O)c1ccc(NC(=S)NC(=O)c2ccco2)cc1\nTarget: O75496, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0044"}
|
| 46 |
+
{"class": "reasoning", "compound_name": "WHI-P131", "compound_smiles": "COc1cc2ncnc(Nc3ccc(O)cc3)c2cc1OC", "compound_inchikey": "HOZUXBLMYUPGPZ-UHFFFAOYSA-N", "target_uniprot": "Q13043", "target_gene": "MST1", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: WHI-P131\nSMILES: COc1cc2ncnc(Nc3ccc(O)cc3)c2cc1OC\nTarget: MST1 (Q13043), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0045"}
|
| 47 |
+
{"class": "reasoning", "compound_name": "SID17511938", "compound_smiles": "COc1ccc(C2Nc3ccccc3C(=O)N2c2ccc(F)cc2)cc1CSc1ccccn1", "compound_inchikey": "UTKZKGGRJXHSHT-UHFFFAOYSA-N", "target_uniprot": "P47898", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: SID17511938\nSMILES: COc1ccc(C2Nc3ccccc3C(=O)N2c2ccc(F)cc2)cc1CSc1ccccn1\nTarget: P47898, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0046"}
|
| 48 |
+
{"class": "reasoning", "compound_name": "APITOLISIB", "compound_smiles": "Cc1c(CN2CCN(C(=O)[C@H](C)O)CC2)sc2c(N3CCOCC3)nc(-c3cnc(N)nc3)nc12", "compound_inchikey": "YOVVNQKCSKSHKT-HNNXBMFYSA-N", "target_uniprot": "Q9Y572", "target_gene": null, "target_family": "protein", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: APITOLISIB\nSMILES: Cc1c(CN2CCN(C(=O)[C@H](C)O)CC2)sc2c(N3CCOCC3)nc(-c3cnc(N)nc3)nc12\nTarget: Q9Y572, protein\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0047"}
|
| 49 |
+
{"class": "reasoning", "compound_name": "ALECTINIB", "compound_smiles": "CCc1cc2c(cc1N1CCC(N3CCOCC3)CC1)C(C)(C)c1[nH]c3cc(C#N)ccc3c1C2=O", "compound_inchikey": "KDGFLJKFZUIJMX-UHFFFAOYSA-N", "target_uniprot": "Q9Y6R4", "target_gene": "MAP3K4", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: ALECTINIB\nSMILES: CCc1cc2c(cc1N1CCC(N3CCOCC3)CC1)C(C)(C)c1[nH]c3cc(C#N)ccc3c1C2=O\nTarget: MAP3K4 (Q9Y6R4), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0048"}
|
| 50 |
+
{"class": "reasoning", "compound_name": "CAPMATINIB", "compound_smiles": "CNC(=O)c1ccc(-c2cnc3ncc(Cc4ccc5ncccc5c4)n3n2)cc1F", "compound_inchikey": "LIOLIMKSCNQPLV-UHFFFAOYSA-N", "target_uniprot": "P53671", "target_gene": "LIMK2", "target_family": "kinase", "num_assays": 2, "num_sources": 2, "evidence_quality": "silver", "context_text": "Compound: CAPMATINIB\nSMILES: CNC(=O)c1ccc(-c2cnc3ncc(Cc4ccc5ncccc5c4)n3n2)cc1F\nTarget: LIMK2 (P53671), kinase\n\nThis compound has been experimentally confirmed as INACTIVE against this target.\n\nExplain the likely molecular and pharmacological reasons for this inactivity.\nConsider: binding site compatibility, selectivity profile, structural features,\nmechanism of action, and any known SAR (structure-activity relationship) data.", "split": "test", "question_id": "L3-0049"}
|
data/llm_benchmarks/l4_tested_untested.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ppi_llm/ppi_l1_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ppi_llm/ppi_l1_metadata.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ppi-l1",
|
| 3 |
+
"created": "2026-03-21T01:04:59.015966+00:00",
|
| 4 |
+
"n_total": 1200,
|
| 5 |
+
"n_per_class": {
|
| 6 |
+
"direct_experimental": 300,
|
| 7 |
+
"systematic_screen": 300,
|
| 8 |
+
"computational_inference": 300,
|
| 9 |
+
"database_absence": 300
|
| 10 |
+
},
|
| 11 |
+
"difficulty_distribution": {
|
| 12 |
+
"easy": "480",
|
| 13 |
+
"medium": "420",
|
| 14 |
+
"hard": "300"
|
| 15 |
+
},
|
| 16 |
+
"split_distribution": {
|
| 17 |
+
"test": "720",
|
| 18 |
+
"fewshot": "240",
|
| 19 |
+
"val": "240"
|
| 20 |
+
},
|
| 21 |
+
"seed": 42
|
| 22 |
+
}
|
data/ppi_llm/ppi_l2_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ppi_llm/ppi_l2_metadata.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ppi-l2",
|
| 3 |
+
"created": "2026-03-21T01:08:26.491226+00:00",
|
| 4 |
+
"n_total": 500,
|
| 5 |
+
"design": "constructed_evidence_fallback",
|
| 6 |
+
"pair_distribution": {
|
| 7 |
+
"1": "200",
|
| 8 |
+
"2": "200",
|
| 9 |
+
"3": "100"
|
| 10 |
+
},
|
| 11 |
+
"split_distribution": {
|
| 12 |
+
"test": "400",
|
| 13 |
+
"fewshot": "50",
|
| 14 |
+
"val": "50"
|
| 15 |
+
},
|
| 16 |
+
"seed": 42
|
| 17 |
+
}
|
data/ppi_llm/ppi_l3_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ppi_llm/ppi_l3_metadata.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ppi-l3",
|
| 3 |
+
"created": "2026-03-22T07:45:41.330745+00:00",
|
| 4 |
+
"n_total": 200,
|
| 5 |
+
"compartment_distribution": {
|
| 6 |
+
"true": "100",
|
| 7 |
+
"false": "100"
|
| 8 |
+
},
|
| 9 |
+
"split_distribution": {
|
| 10 |
+
"test": "160",
|
| 11 |
+
"fewshot": "20",
|
| 12 |
+
"val": "20"
|
| 13 |
+
},
|
| 14 |
+
"source_distribution": {
|
| 15 |
+
"humap": "158",
|
| 16 |
+
"huri": "42"
|
| 17 |
+
},
|
| 18 |
+
"seed": 42
|
| 19 |
+
}
|
data/ppi_llm/ppi_l4_dataset.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
data/ppi_llm/ppi_l4_metadata.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "ppi-l4",
|
| 3 |
+
"created": "2026-03-21T01:14:59.583565+00:00",
|
| 4 |
+
"n_total": 500,
|
| 5 |
+
"n_tested": 250,
|
| 6 |
+
"n_untested": 250,
|
| 7 |
+
"n_tested_pre_2015": 125,
|
| 8 |
+
"n_tested_post_2020": 125,
|
| 9 |
+
"n_untested_trick": 125,
|
| 10 |
+
"n_untested_obvious": 125,
|
| 11 |
+
"split_distribution": {
|
| 12 |
+
"fewshot": 50,
|
| 13 |
+
"val": 50,
|
| 14 |
+
"test": 400
|
| 15 |
+
},
|
| 16 |
+
"seed": 42
|
| 17 |
+
}
|