Fig. 6From: A sparse autoencoder-based deep neural network for protein solvent accessibility and contact number prediction3D visualization of the difference between predicted and observed contact number of the HisC protein (PDB entry: 4wbtA). Different colors represent different numbers of error predicted contact number. The number of errors corresponding to a color is displayed on the right. The closer the blue indicates the more accurate the prediction; otherwise, if it is close to the red, the prediction has more errorsBack to article page