Datasets:

SidraBhatti
/

caspase4-inhibitor-screening-data

Error code:   DatasetGenerationCastError
Exception:    DatasetGenerationCastError
Message:      An error occurred while generating the dataset

All the data files must have the same columns, but at some point there are 6 new columns ({'active', 'drugbank_id', 'token_count', 'truncated', 'pIC50', 'chembl_id'}) and 3 missing columns ({'filename', 'cluster_id', 'score'}).

This happened while the csv dataset builder was generating data using

hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data/chemberta_input_features.csv (at revision f9a32abcd5b328287b46e36ea6d2d2eab9de35df), ['hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/Screening energy VINA_caspase-4.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/chemberta_input_features.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/qsar_data_with_descriptors.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/screening_with_drugbank_ids.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/top10_molecular_properties.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/top10_molecular_properties_with_PK_like_cols.csv']

Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1848, in _prepare_split_single
                  writer.write_table(table)
                  ~~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 765, in write_table
                  self._write_table(pa_table, writer_batch_size=writer_batch_size)
                  ~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 773, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              drugbank_id: string
              chembl_id: string
              token_count: int64
              truncated: bool
              pIC50: double
              active: bool
              -- schema metadata --
              pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 972
              to
              {'filename': Value('string'), 'score': Value('float64'), 'cluster_id': Value('int64')}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      builder, max_dataset_size_bytes=max_dataset_size_bytes
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  ):
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1850, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
                  ...<4 lines>...
                  )
              datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
              
              All the data files must have the same columns, but at some point there are 6 new columns ({'active', 'drugbank_id', 'token_count', 'truncated', 'pIC50', 'chembl_id'}) and 3 missing columns ({'filename', 'cluster_id', 'score'}).
              
              This happened while the csv dataset builder was generating data using
              
              hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data/chemberta_input_features.csv (at revision f9a32abcd5b328287b46e36ea6d2d2eab9de35df), ['hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/Screening energy VINA_caspase-4.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/chemberta_input_features.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/qsar_data_with_descriptors.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/screening_with_drugbank_ids.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/top10_molecular_properties.csv', 'hf://datasets/SidraBhatti/caspase4-inhibitor-screening-data@f9a32abcd5b328287b46e36ea6d2d2eab9de35df/top10_molecular_properties_with_PK_like_cols.csv']
              
              Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

filename string	score float64	cluster_id int64
DB11742_docking_min1.mol2	-10	0
DB00843_docking_min1.mol2	-7.6	0
DB08990_docking_min1.mol2	-6.6	0
DB00645_docking_min1.mol2	-6.5	0
DB11125_docking_min1.mol2	-6.5	0
DB09345_docking_min1.mol2	-6.1	0
DB12554_docking_min1.mol2	-5.9	0
DB01081_docking_min1.mol2	-5.8	0
DB12278_docking_min1.mol2	-5.7	0
DB01146_docking_min1.mol2	-5.6	0
DB00454_docking_min1.mol2	-5.5	0
DB08988_docking_min1.mol2	-5.4	0
DB06684_docking_min1.mol2	-9.4	1
DB00480_docking_min1.mol2	-8.2	1
DB00210_docking_min1.mol2	-9.2	2
DB11071_docking_min1.mol2	-7.6	2
DB13953_docking_min1.mol2	-7.5	2
DB00788_docking_min1.mol2	-7.3	2
DB04575_docking_min1.mol2	-7.2	2
DB04824_docking_min1.mol2	-7.1	2
DB01428_docking_min1.mol2	-6.9	2
DB01241_docking_min1.mol2	-6.4	2
DB11994_docking_min1.mol2	-6.4	2
DB01357_docking_min1.mol2	-6.3	2
DB04573_docking_min1.mol2	-6.3	2
DB00946_docking_min1.mol2	-6.2	2
DB00374_docking_min1.mol2	-6.1	2
DB00573_docking_min1.mol2	-6.1	2
DB11201_docking_min1.mol2	-5.8	2
DB11791_docking_min1.mol2	-9.2	3
DB01544_docking_min1.mol2	-9.1	4
DB01595_docking_min1.mol2	-6.7	4
DB01068_docking_min1.mol2	-6.6	4
DB06589_docking_min1.mol2	-9.1	5
DB00705_docking_min1.mol2	-8.1	5
DB00482_docking_min1.mol2	-7.6	5
DB08881_docking_min1.mol2	-7.6	5
DB00576_docking_min1.mol2	-7.3	5
DB00216_docking_min1.mol2	-7.1	5
DB11817_docking_min1.mol2	-6.9	5
DB00808_docking_min1.mol2	-6.7	5
DB00952_docking_min1.mol2	-6.7	5
DB00918_docking_min1.mol2	-6.4	5
DB01621_docking_min1.mol2	-6.4	5
DB06147_docking_min1.mol2	-6.2	5
DB13165_docking_min1.mol2	-6.1	5
DB01194_docking_min1.mol2	-5.6	5
DB00869_docking_min1.mol2	-5.5	5
DB00669_docking_min1.mol2	-5.3	5
DB00734_docking_min1.mol2	-9	6
DB09195_docking_min1.mol2	-8.8	6
DB00298_docking_min1.mol2	-7.8	6
DB00656_docking_min1.mol2	-7.7	6
DB00972_docking_min1.mol2	-7.7	6
DB01149_docking_min1.mol2	-7.6	6
DB09034_docking_min1.mol2	-7.2	6
DB00402_docking_min1.mol2	-6.3	6
DB01198_docking_min1.mol2	-6.3	6
DB01067_docking_min1.mol2	-9	7
DB09183_docking_min1.mol2	-8.4	7
DB01016_docking_min1.mol2	-8.3	7
DB01582_docking_min1.mol2	-8.1	7
DB01298_docking_min1.mol2	-8	7
DB00880_docking_min1.mol2	-7.9	7
DB00436_docking_min1.mol2	-7.8	7
DB00664_docking_min1.mol2	-7.7	7
DB11362_docking_min1.mol2	-7.7	7
DB00263_docking_min1.mol2	-7.6	7
DB08798_docking_min1.mol2	-7.5	7
DB00524_docking_min1.mol2	-7.4	7
DB01325_docking_min1.mol2	-7.4	7
DB00706_docking_min1.mol2	-7.2	7
DB00359_docking_min1.mol2	-7	7
DB00391_docking_min1.mol2	-7	7
DB01581_docking_min1.mol2	-7	7
DB00469_docking_min1.mol2	-6.9	7
DB00562_docking_min1.mol2	-6.9	7
DB06821_docking_min1.mol2	-6.9	7
DB00862_docking_min1.mol2	-6.8	7
DB00203_docking_min1.mol2	-6.7	7
DB01015_docking_min1.mol2	-6.7	7
DB00276_docking_min1.mol2	-6.6	7
DB00554_docking_min1.mol2	-6.5	7
DB00999_docking_min1.mol2	-6.5	7
DB01382_docking_min1.mol2	-6.5	7
DB00214_docking_min1.mol2	-6.4	7
DB00891_docking_min1.mol2	-6.4	7
DB02925_docking_min1.mol2	-6.3	7
DB01021_docking_min1.mol2	-6.1	7
DB00774_docking_min1.mol2	-6	7
DB01324_docking_min1.mol2	-5.8	7
DB00232_docking_min1.mol2	-5.7	7
DB06729_docking_min1.mol2	-5.7	7
DB01299_docking_min1.mol2	-5.6	7
DB00887_docking_min1.mol2	-5.3	7
DB00872_docking_min1.mol2	-8.9	8
DB12867_docking_min1.mol2	-8.5	8
DB01261_docking_min1.mol2	-8.4	8
DB04908_docking_min1.mol2	-8.4	8
DB01184_docking_min1.mol2	-8.3	8

End of preview.

CASP4 Inhibitor Screening — AI Pipeline Data

Processed data from an AI-accelerated virtual screening pipeline that identified repositionable DrugBank compounds as candidate CASP4 (Caspase-4) inhibitors for Alzheimer's disease, using ChemBERTa embeddings, physicochemical descriptors, and Random Forest classification/regression, followed by molecular docking, MD simulation, and MM/PBSA free-energy validation.

Manuscript status: accepted; citation to be added upon publication. Code: github.com/mubashirhassangcul/Caspase4-inhibitor-screening

Role & Attribution

Sidra Bhatti led the AI-based prediction pipeline: ChemBERTa embedding generation, hybrid feature engineering (768-dim ChemBERTa + 6 RDKit descriptors), and development of the Random Forest classifier (active/inactive) and regressor (pIC50 prediction) used to prioritize candidates. Molecular docking, MD simulations, and MM/PBSA analysis were performed by collaborators on the project (see the GitHub repository for full author list).

Files

File	Description
`Screening energy VINA_caspase-4.csv`	AutoDock Vina docking scores (initial structure-based screen)
`screening_with_drugbank_ids.csv`	Processed screening results mapped to DrugBank IDs
`chemberta_input_features.csv`	Hybrid feature set metadata (ChEMBL ID, token count, pIC50 label, active/inactive) used to train the RF classifier/regressor — the 768-dim ChemBERTa embeddings themselves are derived from each compound's SMILES at feature-generation time, not stored here
`qsar_data_with_descriptors.csv`	QSAR descriptor table for all screened compounds
`top10_molecular_properties.csv`	Molecular properties of the top-10 hits
`top10_molecular_properties_with_PK_like_cols.csv`	Extended PK-relevant properties for the top-10 hits

Note: SMILES/structure columns (smiles, canonical_smiles) and the SMILES-only top10_smiles.csv file have been excluded pending confirmation of DrugBank's redistribution terms for structure data pulled from their database. All rows are keyed by drugbank_id — structures can be re-fetched directly from DrugBank or ChEMBL using that ID.

Pipeline summary

DrugBank compounds → structure-based docking screen (~1,739 hits) → ChEMBL pIC50 labeling (active if pIC50 ≥ 8.2) → ChemBERTa + RDKit hybrid features → Random Forest classifier (774-dim, 500 trees; ~95% accuracy, ROC-AUC = 0.73, ~3.5-fold enrichment) → Random Forest regressor (pIC50 ranking) → top-10 selection → ADMET/PK-PD filtering → docking + MD/MM-PBSA validation of top 5 vs. donepezil reference.

What's excluded

Raw DrugBank data (drugbank.xml) is not included — DrugBank's terms of use restrict redistribution; register at drugbank.ca to reproduce the full screen from source.
SMILES/structure columns are excluded pending license confirmation (see note above under Files).

Citation

Manuscript accepted; BibTeX will be added here once publication details are final.

Downloads last month: 23

Total file size:

248 kB