TY - JOUR T1 - Predicting Negative Control Drugs to Support Research in Drug Safety JF - bioRxiv DO - 10.1101/380832 SP - 380832 AU - Yun Hao AU - Nicholas P. Tatonetti Y1 - 2018/01/01 UR - http://biorxiv.org/content/early/2018/07/30/380832.abstract N2 - The lack of high-quality reference data is a major limitation in drug safety and drug discovery science. Unreliable standards prohibit the use of supervised learning methods and make evaluation of algorithms difficult. While some data is available for positive examples (e.g. which drugs are associated with a side effect), there are no systematic resources of negative controls. To solve this issue, we introduced SIDERctrl, a computational method that ranks drugs based on the likelihood of not causing a side effect. We applied SIDERctrl to predict negative controls from unreported drugs of 890 side effects in SIDER. Our predictions decreased the false negative rate by one-third according to a validation study using AEOLUS data. Three sets of predicted negative controls by different thresholds of precision were provided, and can be accessed at http://tatonettilab.org/resources/negative-drugs.html. This new reference standard will be important in chemical biology, drug development, and pharmacoepidemiology.KEY POINTSThe lack of systematic resources providing negative control drugs limits the performance of existing research in drug safety.We developed a novel method that integrated chemical and biological properties a drug and the target proteins to calculate the likelihood of the drug being negative control.We applied our method to 890 side effects, and showed that our method significantly decreased the false negative rate of predictions. ER -