Skip to main content

Table 6 The selection of datasets on which models were trained, and details pertaining to these sets

From: Building attention and edge message passing neural networks for bioactivity and physical–chemical property prediction

DatasetTasksTypeCompoundsSplitMetric
MUV17Classification93,127RandomPRC-AUC
HIV1Classification41,913ScaffoldROC-AUC
BBBP1Classification2053ScaffoldROC-AUC
Tox2112Classification8014RandomROC-AUC
SIDER27Classification1427RandomROC-AUC
QM812Regression21,786RandomMAE
ESOL1Regression1128RandomRMSE
LIPO1Regression4200RandomRMSE