Classification of DNA Microarrays Using Deep Learning to identify Cell Cycle Regulated Genes

Abstract

The aim of this work is to take advantage of the power and growth of machine learning methods and deep learning algorithms in the biomedical field, and how to use it to predict and recognize repetitive patterns. The ultimate goal is to analyze the large amount set of data produced from the DNA (Deoxyribonucleic acid) microarrays technology. We can use this data to extract facts, information, and skills, such as gene expression level. Our target here is to classify two genes’ types. The first represents cell cycle regulated genes and the second represents the non-cell cycle ones. For the classification purpose, we preprocess the data, and we implement deep learning models. Then we evaluate our approach and compare its precision with Liu and al results. In the literature, the latest approaches are depending on processing the numerical data related to the DNA microarrays genes progression. In our work, we adopt a novel approach using directly the Microarrays image data. We use the Convolutional Neural Network and the fully connected neural network algorithms, to classify our processed image data. The experiments demonstrate that our approach outperforms the state of art by a margin of 20 per cent. Our model accomplishes real time test accuracy of ~ 92.39 % at classifying.