The server sends back the topology prediction for each sequence. For each prediction only the actual single sequence information is used.