To process a similarity query, we use the following steps:
Input is a Graph Query representing the structure of the Protein you want to be searched
Calculating the Graph Spectra of the Graph Query
Determining node on the SOM output layer with smallest distance between weight vector of this node and graph spectra of query graph. Highlight the neighboring areas containing this node and display a window of proteins with high similarity
Figure 4 shows the 3D structure of protein with the PDB access code 1ssx and 2h5c in the same cluster of SOM 2D output layer.