Prediction of Antimicrobial Peptides Using Machine Learning Approach

IntroductionAntimicrobial peptides (AMPs), also called host defense peptides (HDPs), consist of 12 to 100 amino acids that are part of the innate immune system and can be found among all classes of life including bacteria, fungi, plants, invertebrates, and vertebrates.These AMPs have been found to be effective against disease-causing pathogens.Identification of antimicrobial peptides through in vitro and in vivo experiments on large number of peptides is an expensive and time-consuming approach.This study explores machine learning classifiers for predicting antimicrobial peptides (AMPs) using a diverse set of AMPs ( 2638) and non-AMPs (3700).The RF classifier-based model outperformed other models in both internal and external validations.It correctly predicted known AMPs and non-AMPs, with ChargeD2001, PAAC12 (pseudo amino acid composition), and polarity T13 being crucial features in AMPs' antimicrobial activity.The developed RF-based classification model may be useful in designing and predicting novel potential AMPs. Aim & ObjectivesAim: Development of a predictive model for antimicrobial peptides using machine learning approach. Objectives:1.Collection of antimicrobial peptide data from various databases 2. Calculation of sequence based descriptors for the antimicrobial peptides 3. Identification of important descriptors for model development 4. Development and validation of models for predicting antimicrobial peptides Workflow AMP Prediction Prediction of antimicrobial peptides using developed model Model Validation K-fold cross validation like 5-fold and 10-fold cross validation Model Development Machine learning techniques like k-NN, Random Forest, SVM Descriptor calculation Sequence based descriptors Identification of important descriptors Data Collection Databases like CAMP R 3 Literature Methodology 1. Data collection ❑ Finally, a dataset of 6338 peptide sequence obtained from adding negative and positive dataset were used for the model development.