Fig. 2.From: Predicting the hosts of prokaryotic viruses using GCN-based semi-supervised learningCNN model used in HostG. A Train mode. B Encoding mode. T1, T2, and T3 represent the genomes of viruses that will be fed to CNN to update parameters during back propagation in the train mode. E1, E2, and E3 represent the contigs/genomes that will be fed to pre-trained CNN for encoding the sequences into numerical vectors in the encoding modeBack to article page