EPITOPE-PARATOPE RECOGNITION BY KNOWLEDGE BASED CORRELATION MAPPING USING HOPFIELD NETWORK

摘要

The knowledge based correlation mapping (KBCM) is devised here as a unified measure of the information on antigen-antibody complexes available in the Protein Data Banks. This mapping is estimated from a random training sample using a Hopfield Network while representing the epitope and paratope in a complex as 3-dimensional attributed graphs. Identification of paratopes on immunoglobulins against given epitopes is carried out using pattern matching on Hopfield Network with the help of the KBCM.