Identification of Disease Resistant Plant Genes Using Artificial Neural Network

Much like animals have their defenses against disease-causing pathogens, plants have their own mechanisms to identify and defend against pathogenic microorganisms. Much of this mechanism depends upon disease-resistant genes, also known as ‘R’ genes. Early identification of these R genes is essential in any crop improvement program, especially in a time when plant diseases are one of the biggest causes of crop failure worldwide. Existing methods operate on domain dependence which have several drawbacks and can cause new or low similarity sequences to go unrecognized. In this paper, a Machine Learning method, employing a domain-independent approach, was developed and evaluated which improves upon or eliminate the drawbacks of existing methods. Data sets were obtained from publicly accessible repositories, and feature extraction generated 10,049 number of features. Batch Normalization was used on the models, and we were able to achieve a 97% accuracy on the test dataset which is greater than anything else in the literature that uses the same approach.