N-to-1 NN architecture for predicting enzymatic class. An N-to-1 Neural Network. N copies of the network (only 3 represented for simplicity) process all the (overlapping) motifs of a predefined length in a sequence. The vectorial outputs fk of these networks are added up and the resulting feature vector f is input to the network to produce the enzymatic class prediction.
Volpato et al. BMC Bioinformatics 2013 14(Suppl 1):S11 doi:10.1186/1471-2105-14-S1-S11