Repository navigation
Current Datasets
Verified 3 November 2026 using the public dataset API, target metadata, and published models.
The server lists 21 datasets, 63,866 imported dataset records, and 21 published models. Dataset records are not a count of unique registry molecules. All listed datasets were marked MIGRATED and model-eligible, with zero failed/not-processed records in the catalog snapshot.
Dataset presence, model definitions, and publication are separate. In particular, a multi-target dataset does not guarantee a published predictor for every target. Consult the model selector for current prediction availability.
Names below preserve the live catalog, including historical spellings such as “Flucanozole” and “Toxcicity.” Descriptions summarize assay context from the prior wiki and catalog provenance. PubMed references are those recorded by the live catalog; where no reference is recorded, that limitation is explicit. Original-study cohort sizes can differ from imported MolClass subsets.
| ID | Dataset | Imported records | Target properties |
|---|---|---|---|
| 1 | Compound induced Neuronal Stem Cell Differentiation | 1,267 | 1 |
| 2 | Caco2 Penetration | 100 | 1 |
| 3 | Flucanozole Synergizer | 1,043 | 1 |
| 4 | Compound accumulation in C. elegans | 476 | 1 |
| 5 | Toxicity Benchmark | 5,351 | 1 |
| 6 | Ames Categorization | 4,261 | 1 |
| 7 | Blood Brain Barrier | 412 | 1 |
| 8 | P.falciparum Sensitivity | 1,252 | 1 |
| 9 | HSP90 PPI Inhibition | 1,614 | 1 |
| 10 | Chembank Cell Cycle Inhibitor Screen | 435 | 1 |
| 11 | BetaCell Transdifferentiation | 907 | 1 |
| 12 | Xenopus Actin Polymerization | 1,359 | 1 |
| 13 | Auto-fluorescence Prediction | 4,910 | 1 |
| 14 | Thrombin Molecule Binding | 4,876 | 1 |
| 15 | Metabolite Likeness | 982 | 1 |
| 16 | Compound aggregation | 732 | 1 |
| 17 | Liver Toxcicity | 921 | 1 |
| 18 | Antifungal Pump Inhibitors | 10,427 | 1 |
| 19 | Mitochondrial Fusion | 19,869 | 1 |
| 23 | Yeast Bioactive Library Cross-Kingdom | 1,570 | 17 |
| 31 | Microsome Clearance (AZ Hackathon) | 1,102 | 1 |
Chemical-genetic screening data concerning neural stem-cell and neurosphere proliferation. Useful for exploring structure-associated effects in this cellular context; retain the original labels rather than interpreting every hit as neuronal differentiation.
Imported records: 1,267. Source identifier property: __legacy_mol_id.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
Mutual: 1,107; Proliferation: 160 |
1,267 |
Catalog literature reference: PubMed 17417631.
Caco-2 permeability classification data used to relate molecular structure to passage across a cell-monolayer model. The hosted target is categorical; it does not return a continuous permeability measurement.
Imported records: 100. Source identifier property: compound_name.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
app-: 68; app+: 32 |
100 |
Catalog literature reference: PubMed 15446816.
Compounds screened for their ability to potentiate fluconazole activity. This dataset supports antifungal-combination exploration; synergy is assay- and organism-dependent.
Imported records: 1,043. Source identifier property: compound_name.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
No Synergy: 904; Flu.Synergy: 139 |
1,043 |
Catalog literature reference: PubMed 21694716.
Compound accumulation in Caenorhabditis elegans, intended to help identify molecules more likely to reach targets in worm-based screens. Accumulation is distinct from a specific molecular mechanism of action.
Imported records: 476. Source identifier property: external_predictor.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
none: 402; Accumulates: 74 |
476 |
Catalog literature reference: PubMed 20512140.
An Ames-mutagenicity benchmark retained under the general catalog name Toxicity Benchmark. Interpret its Toxic/none labels in the original mutagenicity context, not as a universal toxicity or clinical-safety judgment.
Imported records: 5,351. Source identifier property: cas_no.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
none: 2,754; Toxic: 2,597 |
5,351 |
Catalog literature reference: PubMed 19702240.
Ames test classifications associated with toxicophore-based mutagenicity work. This is a separate dataset from Toxicity Benchmark, with its own imported records and mutagen/nonmutagen labels.
Imported records: 4,261. Source identifier property: cas_no.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
mutagen: 2,366; nonmutagen: 1,895 |
4,261 |
Catalog literature reference: PubMed 15634026.
Blood–brain barrier penetration classifications. The target distinguishes BBB-penetrating and nonpenetrating compounds; it is not a prediction of therapeutic efficacy in the brain.
Imported records: 412. Source identifier property: compound_name.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
BBB+: 273; BBB-: 139 |
412 |
Catalog literature reference: PubMed 16180914.
Plasmodium falciparum screening data concerning differential chemical sensitivity. The imported target distinguishes inhibitory and inactive compounds.
Imported records: 1,252. Source identifier property: pubchem_compound_cid.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
inactive: 906; inhibitor: 346 |
1,252 |
Catalog literature reference: PubMed 19734910.
A screen for disruption of an Hsp90 co-chaperone protein–protein interaction. The model concerns that assay endpoint rather than all possible Hsp90-related activities.
Imported records: 1,614. Source identifier property: pubchem_compound_cid.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
Hsp90_ppi-: 1,017; Hsp90_ppi+: 597 |
1,614 |
Catalog literature reference: not recorded.
ChemBank cell-cycle screening data, historically described as a screen for modulators of phosphonucleolin. The class labels encode the imported cell-cycle assay outcome.
Imported records: 435. Source identifier property: chembankid.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
cellcycle-: 277; cellcycle+: 158 |
435 |
Catalog literature reference: not recorded.
ChemBank screening data for small molecules associated with a beta-cell-like phenotype, using insulin expression in an alpha-cell context. The endpoint is an experimental phenotype, not demonstrated diabetes treatment.
Imported records: 907. Source identifier property: chembankid.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
betacelltrans-: 490; betacelltrans+: 417 |
907 |
Catalog literature reference: not recorded.
Screening for inhibition of actin polymerization in PIP2-stimulated Xenopus egg extracts. Useful for exploring chemical effects on the actin-assembly assay.
Imported records: 1,359. Source identifier property: chembankid.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
actin-: 997; actin+: 362 |
1,359 |
Catalog literature reference: not recorded.
ChemBank-derived autofluorescence annotations. These can help identify compounds whose own fluorescence may interfere with fluorescence-based screening readouts.
Imported records: 4,910. Source identifier property: chembankid.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
mutual: 3,978; Autofluorescence: 932 |
4,910 |
Catalog literature reference: not recorded.
ChEMBL-derived thrombin activity/binding classifications. The hosted model target is categorical Active/Inactive; it does not estimate a binding constant.
Imported records: 4,876. Source identifier property: chembl_id.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
Active: 3,610; Inactive: 1,266 |
4,876 |
Catalog literature reference: not recorded.
A metabolite-likeness dataset separating HMDB and ZINC source classes. The source labels support comparison with metabolite-like chemical space; they do not prove that a candidate is an endogenous metabolite.
Imported records: 982. Source identifier property: compound_name.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
ZINC: 504; HMDB: 478 |
982 |
Catalog literature reference: PubMed 22194963.
Colloidal-aggregation annotations for small molecules. Useful for investigating a potential source of assay artifacts and nonspecific apparent activity.
Imported records: 732. Source identifier property: compound_name.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
Aggregator: 493; Non-aggregator: 239 |
732 |
Catalog literature reference: PubMed 22625864.
Literature-derived reports of liver toxicity. The distinction is toxicity reported versus no toxicity reported; absence of a report is not evidence that a molecule is safe.
Imported records: 921. Source identifier property: pubchem_substance_id.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
toxicity reported: 628; no toxicity reported: 293 |
921 |
Catalog literature reference: PubMed 20014752.
Antifungal efflux-pump-related screening data. The imported filename identifies PubChem AID 485275. Interpret activity in the source assay rather than assuming a validated pump-binding mechanism.
Imported records: 10,427. Source identifier property: pubchem_substance_id.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
inactive: 9,000; active: 1,427 |
10,427 |
Catalog literature reference: not recorded. Assay provenance: PubChem AID 485275 (identified by the imported filename).
A yeast-model screen for compounds affecting mitochondrial fusion. The imported filename identifies PubChem AID 1362; the categorical labels reflect the screening endpoint.
Imported records: 19,869. Source identifier property: pubchem_substance_id.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
class |
inactive: 15,974; active: 3,895 |
19,869 |
Catalog literature reference: not recorded. Assay provenance: PubChem AID 1362 (identified by the imported filename).
A cross-kingdom bioactivity library with 17 target properties covering multiple organism and cellular contexts. Each endpoint has its own class support; some targets have missing labels. The API exposes target codes rather than a complete assay-code dictionary, so the codes below are preserved without guessing expanded names.
Imported records: 1,570. Source identifier property: compound_name.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
msp_at |
non-active: 919; active: 651 |
1,570 |
msp_ce |
non-active: 1,389; active: 181 |
1,570 |
msp_ce_emb |
non-active: 1,389; active: 12 |
1,401 |
msp_ce_gro |
non-active: 1,389; active: 134 |
1,523 |
msp_ce_nema |
non-active: 1,389; active: 21 |
1,410 |
msp_dr |
non-active: 1,174; active: 396 |
1,570 |
msp_ec_mcm |
0: 1,117; 1: 453 |
1,570 |
msp_ec_uoe |
0: 1,398; 1: 172 |
1,570 |
msp_hela |
non-active: 987; active: 583 |
1,570 |
msp_hela_cells |
non-active: 987; active: 196 |
1,183 |
msp_hela_dna |
non-active: 987; active: 397 |
1,384 |
msp_ot |
active: 788; non-active: 782 |
1,570 |
msp_ot_lp |
non-active: 782; active: 47 |
829 |
msp_ot_sc |
non-active: 782; active: 153 |
935 |
msp_sa |
non-active: 1,141; active: 429 |
1,570 |
msp_sc |
non-active: 1,109; active: 461 |
1,570 |
msp_sp |
non-active: 1,024; active: 546 |
1,570 |
Catalog literature reference: not recorded.
AstraZeneca hackathon data with categorical human microsomal intrinsic-clearance annotations. These four classes summarize metabolic-clearance categories rather than a continuous clearance rate. The imported filename identifies the human microsome training set.
Imported records: 1,102. Source identifier property: vendor_id.
| Target property | Class labels and record counts | Labeled records |
|---|---|---|
classifier |
medium: 291; fast: 286; slow: 275; instant: 250 |
1,102 |
Catalog literature reference: not recorded.
Class support counts describe imported nonblank target labels before feature exclusions or training splits. They need not equal a published build's training, validation, or holdout totals. Do not fill missing target labels with a negative class. For dataset 23, numeric labels 0 and 1 are shown exactly as returned; their assay-specific interpretation must come from the dataset provenance.
Retrieve /api/v1/datasets with pagination until all total entries are covered. Read /api/v1/model-datasets/{datasetId} for targets and support counts. Compare with /api/v3/models for publication availability. Update this page's verification date and all totals together; never infer hosted datasets from benchmark files, local databases, or historical wiki counts.
Wiki home · Online MolClass · Current datasets · Repository documentation
Documentation reviewed 3 November 2026. Live catalog counts and deployment state can change.