All machine learning models used in this tool have been trained using the datasets in this page. Their binary features indicate the presence/absence of a given annotation within the targets associated with each example compound, and the class label indicates whether the compound has been associated with mouse longevity ('1') or not ('0'). The description files include a list of all features in the data and the class label definition.

Other files available for download:

  • A source file of all human proteins that includes their names and STRING IDs. These are valid inputs for the Target Prediction functionality.
  • A list of all compounds for which there is known evidence for their class label value (for male mice, female mice, or both sexes). Note: the class labels are created from evidence of peer-reviewed studies, with information from DrugAge, as described in the source paper.

Dataset Instances/Features Filesize Description File
Targets - Neighbour Enrichment (all categories).tsv 143/6460 1,874 KB DescriptionFile - NE All Categories dataset.tsv
FA Targets (InterPro Domains).tsv 136/476 135 KB DescriptionFile - FA InterPro dataset.tsv
NE Targets (GO Cellular Component).tsv 134/363 103 KB DescriptionFile - NE Component dataset.tsv
NE Targets (GO Biological Process).tsv 143/2570 749 KB DescriptionFile - NE Process dataset.tsv
NE Targets (KEGG Pathways).tsv 143/298 90 KB DescriptionFile - NE KEGG dataset.tsv
NE Targets (Reactome Pathways).tsv 143/1050 309 KB DescriptionFile - NE Reactome dataset.tsv
NE Targets (WikiPathways).tsv 140/598 171 KB DescriptionFile - NE WikiPathways dataset.tsv
MM Molecular Fingerprints dataset.tsv 158/397 133 KB Data Dictionary - Molecular Fingerprints

Other files: List of human proteins names and STRING IDs.tsv List of compounds with assigned class labels.tsv