Protein Secondary Structure Prediction Without Alignment Using Graph Neural Network

Complex molecules known as proteins carry out a number of essential tasks in the human body. Protein structure are interwoven and determine each protein’s particular activity. Graph Neural Net (GNN) has developed as an effective deep learning method to extract information from protein structures, which may be represented by graphs of amino acid residues. This paper suggests utilizing a graph neural network to predict protein structure from amino acid sequences without alignment. In this instance, nodes (amino acids) and connecting edges (distances between amino acids) may be used to instantiate a protein in a network. To demonstrate the scalability of the suggested technique, many experiments are carried out using various benchmark datasets, including RS126 and RSCB PDB. The simulation results show that in the majority of cases for different data sets, the recommended strategy outperforms other comparable methods.