1st International and 10th National Iranian Conference on Bioinformatics
Home Accepted Papers
Mutation prediction of Infectious viruses based on Different Machine learning approaches
Paper ID : 1439-ICB10
Authors:
Khashayar Ehteshami1, Mohammad Hadi Azarabad2, paria pashootan3, Negar Khalili3, Fereshteh Fallah3, Seyed AmirAli Ghafourian Ghahramani4, Kaveh Kavousi *3
1Department of Computer Engineering, School of Science and Engineering, Sharif University of Technology International campus-kish island
2University of tehran
3University of Tehran
4Assistant Professor of Computer Science, Department of Computer Engineering, Sharif University of Technology, International Campus, Kish Island
Abstract:
In general, the ability to predict the evolution of a pathogen enhances our ability to control, prevent, and treat diseases. Usually, only mutations that can escape the host immune system and affect the severity can sustain and spread throughout generations. Several different pandemics have happened through the years. For example, the 1918 influenza pandemic was one of the most severe in recent history, and the H1N1 virus caused it. By mutation prediction, pandemics can be recognized before they happen. Variational AutoEncoders (VAEs) and Generative Adversarial Networks (GANs) generate new samples from our data in machine learning. Sequence-to-Sequence (Seq2seq) networks are primarily used in translation tasks to generate a new sample from the previous one. The method that we use is a combination of GAN networks and sequence to sequence networks. We create a Seq2seq network as a Generator of our model with the help of Long Short Term Memories (LSTMs) and then use a discriminator to distinguish whether the sequences are fake or real. The most challenging task is that GANs are not good in sequential data, and Seq2seq has some problems in the long length of sequences. For the result, we find out which sequence is more possible for the mutant in the future, and we can use these results for preventing a future pandemic.
Keywords:
Machine Learning; GANs; Deep Learning; Sequence Generating
Status : Paper Accepted (Poster Presentation)