ChAFF datasets

This dataset collection contains ~200K curated Active compound lists from ~90 different BioAssay datasets, focusing on known assay interference artifacts. We applied SMILES standardization using RDKit and MolVS, including molecule sanitization and fragment removal. The final dataset is suitable for training and evaluating machine learning models.

Types and Number of Active Compounds

Type	NumActiveCompounds
Absorbance	1486
Artifact	10952
Autofluoresence	32054
ColloidalAggregators	19533
HeavyHitters	71981
LuciferaseInhibition	32831
Misannotation	39
Reactivity	3107
REDOX	217

Dataset Columns

Column	Description
Type	Task domain (e.g. Absorbance)
DatasetName	Source dataset name
AID	Pubchem Assay ID
ID	Identifier for the compound
IDType	Type of identifier (e.g. CID)
SMILES	Curated SMILES

Datasets can be found in the data folder.

Dataset summary

A summary file is uploaded, which lists:

• Type
• DatasetName
• AID
• NumActiveCompounds
• PaperTitle
• Reference
• URL
• AssayName
• Description

Dataset summary file can be found: ChAFF_dataset_summary.json

License

Each dataset comes from different sources (i.e., PubChem, Papers). Please check our dataset summary file if you are looking for references.

Usage

Load a dataset in python

Each subset can be loaded into python using the Huggingface datasets library. First, from the command line install the datasets library

$ pip install datasets

then, from within python load the datasets library.

>>> import datasets
>>> from datasets import load_dataset, Features, Value

Specifiy column types to prevent pyarrow error.

features = Features({
    "Type": Value("string"),
    "DatasetName": Value("string"),
    "AID": Value("string"), # Treat int as string
    "ID": Value("string"),
    "IDType": Value("string"),
    "SMILES": Value("string")
})

Now load one of the 'ChAFF' datasets, e.g.,

>>> dataset = datasets.load_dataset("maomlab/ChAFF", name = "default", data_files = "data/Absorbance.csv", split = "train", features = features)

You can modify "data/Absorbance.csv" based on your interest (e.g., "data/Reactivity.csv"). The default is split = "train" as we did not split the datasets.