METHODS OF CLUSTERING GENE AND PROTEIN SEQUENCES

摘要

The invention relates to methods for clustering gene and protein sequences. Inparticular, it involves generationof networks of sequences where the interconnections are based upon a measureof similarity. The invention also provides methodsof optimizing and improving the networks by re-wiring of the network basedupon overlap of the nearest neighbors of given pairsof nodes. The invention further provides methods of identifying clusters ofsequences within the networks and the optimizednetworks based upon the topology of the network. The clusters identifiedrepresent groups of sequences that are related by functionand/or evolution. The invention has particular applicability in annotation ofsequences in databases and identification of functionalhomologs which can be very useful for novel therapeutic and diagnostic targetsbased upon such targets belonging to a cluster orfamily that contains a known sequence such as a diagnostic sequence, antigenor other therapeutic target.