Chromosome Polarity Determination Based on the Total Length and Centromere Location Using Machine Learning Algorithms

Abstract

In this work we determine chromosome polarity based on three machine learning methods: multilayer perceptron (MLP) neural networks, k-nearest neighbor (k-nn) method and support vector machines (SVM). In all three machine learning methods only two chromosome features, total length of the chromosome and the cetromere location, were used to determine the chromosome polarity.  Classification results obtained are 95.94%, 95.255%, and 95.88% for MLP neural networks, k-nn method and SVM respectively.