MolData, a molecular benchmark for disease and target based machine learning

Table 1 Data

PubChem Source	AID count	Active data points	Total data points	% Active datapoints	Unique active molecules	Total unique molecules	% Unique active molecules
Broad Institute	67	125,627	22.2 m	0.56%	85,579	472,858	18.1%
Burnham Center for Chemical Genomics	67	139,021	21.9 m	0.63%	77,159	381,794	20.21%
Emory University Molecular Libraries Screening Center	12	24,195	2.47 m	0.98%	20,964	348,231	6.02%
ICCB-Longwood Screening Facility, Harvard Medical School	11	8358	2.1 m	0.39%	6656	564,021	1.18%
Johns Hopkins Ion Channel Center	22	48,545	6.8 m	0.71%	35,487	344,497	10.30%
NMMLSC	42	48,186	11.5 m	0.42%	37,949	369,431	10.27%
National Center for Advancing Translational Sciences (NCATS)	174	720,319	53.4 m	1.35%	240,096	592,616	40.51%
The Scripps Research Institute Molecular Screening Center	148	275,224	47.6 m	0.58%	142,055	920,418	15.43%
Tox21	57	21,475	0.47 m	5.67%	4183	8743	47.84%

ISSN: 1758-2946