Figure 1From: Predicting receptor-ligand pairs through kernel learningWork flow of the combined kernel classifier. For each candidate, Data was partitioned into training and validation sets and parameters were tuned using a 5 fold validation strategy. The final output of the classifiers was achieved by a leave one out strategy. The classifier values were combined for a final result and a threshold was applied to determine which values are predicted edges in the receptor-ligand bipartite graph.Back to article page