Prediction of novel biochemical class disease related proteins and microRNAs by machine learning approach

Abstract

Proteins and functional RNAs are important components of biological organisms, which play essential roles in biological systems. The identification of proteins and functional RNAs is therefore of great importance for understanding biological processes, discovering new therapeutic targets, and accelerating drug development. Due to the limitation of experimental approaches, various computational tools have been developed to facilitate their identification. In this study, we explored machine learning methods, including decision tree, probabilistic neural networks, k nearest neighbors and support vector machines, to develop prediction systems for multifunctional enzymes, disease related proteins and microRNAs from sequence derived properties irrespective of sequence similarity. The results suggest that machine learning is a promising tool for prediction of disease related proteins and functional RNAs, especially for those with poor sequence similarity to known ones.