Classification and identification of cattle antimicrobial peptides using Artificial Neural Network methodology

Abstract

With the advent of machine learning techniques, a large number of biological problems have been given a solution. The interpretation of massive genomic data is a big challenge to the researchers, but literature shows many computational approaches to counter such problems. Out of many biological issues, one is regarding the Antimicrobial peptides (AMPs), which are the hosts’ defence molecules gaining extensive research attention worldwide. Today, resistance to chemical antibiotics is an unsolved and growing problem. AMPs may be a natural alternative to chemical antibiotics and a potential area of research under applied biotechnology. The present work shows application of Artificial Neural Networks (ANN), a machine learning algorithm on bovine AMPs for prediction. Total of 99 AMPs related to cattle collected from various databases and published literature were taken into study. N-terminal residues, C-terminal residues and full sequences were used for model development and identification (prediction). For N-terminal residues, MultiLayer Perceptron (MLP 31-19-2) was found to be the best model with accuracy 94% while for C-terminal residues and full sequence, MLP 31-14-2 and MLP 31- 16-2 were the best models with accuracies 94% and 92%, respectively for classification of bovine AMPs. The computational approach for AMPs identification from this study may be used to design potent peptides against microbial pathogens.