Protein secondary structure prediction by using a hybrid approach with sliding window concept
Prediction of the secondary structure of proteins is an incredibly imperative conception of bioinformatics. This concept provides a path to solve the crucial problem named protein folding. When we will dig up the middle step of protein folding, it provides help to improve the problem of the medical diagnostic system. There are massive numbers of manners and miscellaneous machine learning algorithms and tools are present to solve this problem for enhanced drug designing. In this paper, pertain the binary encoding scheme applied on primary amino acid sequences with the concept of sliding windowpane on it and define multiple hidden layers and use a hybrid approach that is convolution neural network and multilayer feed-forward network for achieve the maximum accuracy and minimize the error of secondary structure prediction. Collect the database from protein data bank after that trained the network by using the concept of Back-propagation algorithm and Keras. This paper provides the optimized results with the ceiling the accuracy for enhanced drug discovery.