Fig. 2From: A multi-task CNN learning model for taxonomic assignment of human virusesOverview of the MT-CNN/Bayesian pipeline. The pipeline incorporates a sliding window for splitting input sequences into 50-mers for prediction by the MT-CNN model. The outputs of the model are ensembled to identify the number of read and genomic coverage per species in the sample. Finally, both parameters are used to estimate a taxonomic ranking using a naïve Bayes networkBack to article page