Predicting secondary structure of a protein using MIRA algorithm based on tetrapeptide structural words

摘要

The secondary structure prediction (S_ecSP) of a protein has a significant performance for the tertiary structure model in Bioinformatics. The formation of the regions of coil (c)-sheets in the secondary structure (S_ecS_t) of a protein, comprises of long-term of associations among residues. Specifically, the prediction of a random coil chemical shift is generally used to find the secondary structure elements in a folded state and disclose the elements in an unfolded state of a protein. The identification of these regions may lead to the vital biological analysis. Although numerous algorithms are available and worked with very large datasets for S_ecSP, the precision level is not in the upper limit. Thus, in this paper, a Margin Infused Relaxed Algorithm (MIRA) is employed to anticipate the S_ecSP of a protein by increasing the diversity of structural words by improving the accuracy limit. Also, the binomial distribution is employed to optimize tetrapeptide structural words. Benchmark datasets such as CullPDB and CB513 are used in this work to analyze the projected approach. By employing the 10-fold cross validation method, the overall precision of the projected approach is 90.76% in projecting the SecSP of a protein and the prediction accuracy of a random coil is 88.45%. The proposed MIRA algorithm was compared with well-known 9 existing approaches. The final accuracy results illustrated that the projected algorithm provides improved results than the other popular existing algorithms.