An improved Hybrid Neuro Fuzzy Genetic System (I-HNFGS) for protein secondary structure prediction from amino acid sequence

This paper proposes few improvements to “Hybrid Neuro Fuzzy Genetic System (HNFGS)”, which we have implemented and presented in [13] for protein secondary structure prediction. The hybridization of artificial neural networks, genetic algorithms, and fuzzy logic can produce robust solutions for complex prediction problems, such as protein secondary structure prediction. Due to the complex and dynamic nature of biological data, we have proposed a two step process to model protein secondary structure prediction. As the protein secondary structure prediction problem involves a huge number of inputs, the input variables of I-HNGS are selected carefully in the first phase. In the second phase genetic algorithms are used to optimize fuzzy set definition and the shape and type of fuzzy membership functions. The improved HNFGS has produced better prediction results when experimented on three-class (alpha-helix, beta-sheet or coil) protein secondary structure prediction from amino acid sequence. The experimental results indicate that the proposed system has the advantages of high precision, good generalization, and comprehensibility. This system also exhibits the property of rapid convergence in fuzzy rule generation.