"Developing models for the prediction of microbial biotransformation pathways and half-lives of trace organic contaminants in different environments requires as training data easily accessible and sufficiently large collections of respective biotransformation data that are annotated with metadata on study conditions."