The document discusses using drug side effects to identify new drug targets. It describes how analyzing the similarity between side effect profiles of different drugs can reveal shared targets, even for drugs that are chemically dissimilar. The author and others developed databases of drug side effect information from package inserts and text mining. They used this information to build a drug-drug network and test predictions, finding binding or activity for the majority of drug pairs examined. Future work involves better linking side effects to specific targets and direct target prediction.