Described in this blog post is a prototype deep neural network model for predicting the genome sequence of the most prevalent SARS-CoV2 mutant variant.
The various mutant SARS-CoV2 variants' genomic data were sourced from the Nucleotide database of the National Library of Medicine (NLM).
The deep neural network prototyping was done using Python and Keras recurrent neural network API.
As newer mutant variants of SARS-CoV2 continue to infect the world, the prevalence of the SARS-CoV2 mutant variants among the infected have changed periodically during the course of the COVID19 pandemic and will continue to undergo these changes in the future.
By using the knowledge of future changes to the SARS-CoV2 genome, a proactive management strategy for COVID19 can be implemented. A better understanding of the mutation dynamics of SARS-CoV2 can potentially help solve the problems of vaccine escape, immune escape and therapeutic resistance more effectively.
This proactive management strategy for COVID19 has the potential to be an important additional tool that will help reduce the level of SARS-CoV2 infections from a pandemic to just an endemic.
Traditionally, the genomic mutations are modeled at the host level. These models use the mutation rate of the organism in question as one of the key inputs.
Techniques such as Markov Chain Monte Carlo (MCMC) simulations are performed based on the mutation rate and also the probability of mutations in specific regions of the genome in question.
Multiple experiments are performed using these techniques to estimate the genotype as well as the phenotype of the future mutant variants.
A key limitation of the host-level genomic modeling is the huge amount of computational resources needed to run these massively parallel mutation modeling experiments.
Also, this technique does not provide adequate insights into the key genomic regions that are conserved among the most prevalent SARS-CoV2 mutant variants.
By approximating the mutation dynamics of the most prevalent SARS-CoV2 variants at the population level, we are able to observe a more holistic, population level view of how mutant variants are influencing the course of the pandemic.
Using an efficient approximation algorithm such as a deep neural network, the computational requirements for developing these models are also significantly lower than the brute force approaches such as massively parallel MCMC simulations.
The insight into the conserved genomic regions among the most prevalent SARS-CoV2 variants, is an important added benefit of this population based approach to model the SARS-CoV2 mutations. These insights are potentially beneficial in identifying the newer therapeutic and vaccine targets against SARS-CoV2.
Moad Computer is an actionable insights firm. We provide enterprises with end-to-end artificial intelligence solutions. Actionable Insights blog is a quick overview of things we are most excited about.