Figure 5.

Conceptual visualization of the final model built using Weka 3.8.5 MultilayerPerceptron classifier for decision making, relating attributes of virus lineages to the evidence for impact on immunity {no, neutralization, escape}. For simplicity only selected 10 mutations are shown in the input layer of neuronal networks. Circles represent neurons and lines represent synapses. The model contains the input layer of considered attributes (32 mutations, time of the first detection, the total change of charge in S, the charge change in RBD, and the charge change in the S1/S2 cleavage area), one hidden layer to process the classification decision, and the output layer of classes for impact on immunity. Synapses take the input and multiply it by a fitted weight. Neurons add the outputs from all synapses and apply a sigmoid transfer function with a fitted threshold.