Estimation of protein structures by classification of angles between α-carbons of amino acids based on artificial neural networks

Authors:
Ali Karci;Murat Demir
Affiliations:
Fırat University, Department of Computer Engineering, 23119 Elazig, Turkey;Fırat University, Muş Vocational School, Muş, Turkey
Venue:
Expert Systems with Applications: An International Journal
Year:
2009

Citing 2
Cited 2

Hybrid Protein Model (HPM): A Method to Compact Protein 3D-structure Information and Physicochemical Properties

SPIRE '00 Proceedings of the Seventh International Symposium on String Processing Information Retrieval (SPIRE'00)
Protein structure classification by structural transformatio

IJSIS '96 Proceedings of the 1996 IEEE International Joint Symposia on Intelligence and Systems

A3N: An artificial neural network n-gram-based method to approximate 3-D polypeptides structure prediction

Expert Systems with Applications: An International Journal
A novel hybrid classification model of artificial neural networks and multiple linear regression models

Expert Systems with Applications: An International Journal

Quantified Score

Hi-index	12.05

Visualization

Abstract

The protein structures are important for protein functionality. There are a lot of researchers who deal with the investigation of protein structures. In this paper, we studied on the estimation of protein secondary structures based on machine learning. In order to use neural networks for learning angle values between @a-carbons of amino acids, we classified the values of angles. For this study, we used amino acid primary structure chains. The angles of centre carbons in an amino acid sequence in two dimensions were used by grouping the centre carbons three by three. In order to accomplish this study, we used the actual angle values obtained from the data bank for training the artificial neural networks. Then trained networks were used for estimating the angle values between centre carbons in amino acids of the three samples. The success rates are 70%, 68.42%, and 49.12% in the case of the classified angle values, and the success rates are 99.5%, 99.1%, and 98.1% in the case of obtained angle values. The success rates with error less than or equal to 0.8 are 100%, 89.5%, and 77.2% in the case of the classified angle values, and the success rates with error less than or equal to 0.5 are 70%, 82.5% and 73.7%.