Fig. 6: Benchmarking of different machine learning algorithms for paired heavy-light chain antibody sequence classification and identification of potential dengue neutralizing antibody binders.
From: The dengue-specific immune response and antibody identification with machine learning
a Graphical representation of the computational pipeline. b From top left to bottom right: number of paired heavy and light chain CDR3 sequences in training datasets; accuracy of random forest algorithm trained with 200, 150, 100 and 50 decision trees; precision, recall, F1 score and accuracy comparison between random forest (RF) with 200 trees, support vector machines (SVM) and multilayer perceptron (MLP). c Receiver operating curve (ROC) of random forest with 200 trees, support vector machine and multilayer perceptron. d Top identified 20 clones in dengue datasets, predicted class (non-dengue or dengue-specific), mouse eliciting the clone and antigen.
