Published online 03 May 2014 | David Gfeller, Aurélien Grosdidier, Matthias Wirth, Antoine Daina, Olivier Michielin, and Vincent Zoete
SwissTargetPrediction is a web server designed to predict the targets of bioactive small molecules, such as drugs or metabolites, by combining 2D and 3D similarity measures with known ligands. The server can predict targets in five different organisms and allows users to map predictions by homology within and between species. The method leverages the observation that similar bioactive molecules are likely to share similar targets. The predictions are ranked based on a combination of 2D and 3D similarity values, with a probability derived from cross-validation studies. The server provides an intuitive interface for inputting query molecules and selecting the organism for predictions. It also includes homology-based predictions, which expand the list of predicted targets by considering orthologous and paralogous proteins. The accuracy of the predictions is validated using a benchmark dataset, achieving an average AUC value of 0.87. SwissTargetPrediction is freely accessible and aims to facilitate the understanding of molecular mechanisms, predict side effects, and optimize compound design.SwissTargetPrediction is a web server designed to predict the targets of bioactive small molecules, such as drugs or metabolites, by combining 2D and 3D similarity measures with known ligands. The server can predict targets in five different organisms and allows users to map predictions by homology within and between species. The method leverages the observation that similar bioactive molecules are likely to share similar targets. The predictions are ranked based on a combination of 2D and 3D similarity values, with a probability derived from cross-validation studies. The server provides an intuitive interface for inputting query molecules and selecting the organism for predictions. It also includes homology-based predictions, which expand the list of predicted targets by considering orthologous and paralogous proteins. The accuracy of the predictions is validated using a benchmark dataset, achieving an average AUC value of 0.87. SwissTargetPrediction is freely accessible and aims to facilitate the understanding of molecular mechanisms, predict side effects, and optimize compound design.