Small Peptides Prediction Using Machine Learning Methods

Abstract

Small peptides in plants are important signal molecules that trigger cell to cell communication and are crucial for plant growth and defense processes. The unique categories of the small peptides can participate differentially in plant growth and development. Research on the classification of the different categories of small peptides is becoming a challenge for the bioinformatics domain. However, the data classification of the critical datasets including various bioinformatics applications has been smartly tackled by Machine Learning in a few of the researches. This research has been done in three phases. In the first phase, an experiment on different feature selection methods is employed on small peptides data to get the most informative predictors/features so that optimum accuracy with a smaller number of features may be obtained. In the second phase, feature subset selector methods have been used to reduce the dimensions of the data and with a small subset of informative features and a robust and rapid small peptides prediction model has been developed. In the final phase, at first, all the state-of-art machine learning classifiers are taken for a comparative analysis on the small peptides data. Further, this research proposed a novel multi-class random forest (MCRF) classifier to detect the different categories of small peptides. newline