Species specific approach to the development of web-based antimicrobial peptides prediction tool for cattle

Antimicrobial peptides (AMPs) are the defence molecules of the host gaining extensive attention worldwide as these are natural alternative to chemical antibiotics. Machine learning techniques have capabilities to analyse large biological data for detection of hidden pattern in understanding complex underlying biological problems. Presently, development of resistance to chemical antibiotics in cattle is unsolved and growing problem which needs immediate attention. In the present study, attempt was made to apply machine learning algorithms such as Artificial Neuron Network (ANN) and Support Vector Machine (SVM). It was found that performance of SVM based models for in silico prediction/identification of AMPs of cattle is superior than ANN. A total of 99 AMPs related to cattle collected from various databases and published literature were taken for this study. N-terminus residues, C-terminus residues and full sequences were used for model development and identification/prediction. It was found that best SVM models in this case for C-terminus residues, N-terminus residues and full sequence were with kernels Radial Basis Function (RBF), Sigmoid and RBF with accuracy as 95%, 99% and 97%, respectively. These SVM models were implemented on web server and made available to users at http://cabin.iasri.res.in/amp/ for classification/prediction of novel AMPs of cattle. This computational server can accelerate novel AMP discovery from whole genome proteins of a given cattle species for bulk discovery with very high accuracy. This is the first successful attempt for development of species specific approach for prediction/classification of AMPs, which may be used further as a model in other species as well.