Comparing Machine Learning Models for Intron Classification

Abstract

This work explores using Probabilistic Context Free Grammars and Artificial Neural Networks as possible machine learning models for classifying introns into major and minor introns. It presents an intron classification framework that combines probabilistic context free grammars and support vector machines. It also assesses the computational prediction power of these two models in comparison to the Position Weight Matrices technique, which is currently the exclusively used model for intron classification. The comparison is done through experimental analysis, and it shows promising results for Probabilistic Context Free Grammars and Artificial Neural Networks.