Homology Modeling of Myoglobin using Adaptive Neurofuzzy Systems

The problem of nonlinear system identification as applied to protein folding problem is discussed in this paper. The mapping of amino acid sequence and the atomic coordinates of the alpha carbon atoms of the amino acids in a protein is typically a nonlinear problem. System identification is performed using adaptive neurofuzzy techniques. Various ANFIS models are created and the model with the least error is selected. The protein ‘Human Myoglobin Mutant (PDB Id: 2MM1)’ and its homologue ‘Pig Metmyoglobin (PDB Id: 1MYH)’ have been used for the creation and training of the ANFIS model. The tertiary structure of 3 proteins ‘Myoglobin (Horse Heart) wild type complexed with nitrosoethane (PDB Id: 1NPG)’, ‘Loggerhead sea turtle Myoglobin (PDB Id: 1LHS)’ and ‘MetMyoglobin from Yellowfin Tuna (PDB Id: 1MYT)’ have been predicted using the same ANFIS model. It is found that the root mean square errors in the prediction of the tertiary structures of the 3 proteins considered in this study are 4.32, 3.04 and 3.00 respectively.