Here we assess the performance of E3FP relative to ECFP4, a widely used 2D fingerprint. We do so using k-fold cross-validation on ChEMBL20 using various classifiers. See directories for instructions specific to SEA or other classifiers:
sea: k-fold cross-validation of various flavors of ECFP4 and E3FP using the Similarity Ensemble Approach (SEA) as classifier.classifiers: validation using four classifiers: Naive Bayes (NB), Random Forests (RF), Support Vector Machine with a linear kernel (LinSVM), and Artificial Neural Network (NN).classifiers_mean: validation using the four classifiers listed above, but combining all bit fingerprints for a molecule into a single 'float' fingerprint by averaging.