Fig. 5From: DrugShot: querying biomedical search terms to retrieve prioritized lists of small moleculesViolin plots of average precision score distributions for each collection of search terms for each prediction matrix compared with random shuffles of the prediction matrix. Average precision scores for each term were determined based on the rankings of the unweighted drug set created from AutoRIF in each prediction matrix. A AutoRIF literature co-mentions prediction matrix. B DrugRIF literature co-mentions prediction matrix. C Signature similarity prediction matrixBack to article page