A mutation prediction algorithm built to map out subsequent mutations of a strain of HIV virus in a person’s bloodstream over n timesteps, and can be further used in optimizing antiretroviral treatment (allocating the right medication at the right time to avoid drug resistance).
Applied mutations (addition, substitution, etc) through Monte Carlo upon listed initial strain sequences (source: Stanford HIV database); built adversarial network to generate adversarial sequences, and discriminator/classification network to identify valid subsequent sequences to prune MC-mutations
Further contribution of providing a mutation prediction algorithm is classifying HIV antiretroviral medication for specific strains of HIV, thus optimizing medication intake for patients in terms of viral drig resistance and elimination of virus
The above is a plot of the initial strains of HIV in one’s bloodstream, and its simulated strains converged into the most likley possible strains in the next timestep.
HIV is currently a non-curable yet treatable virus that replicates by encoding its own DNA into white blood cells. The virus prevents white blood cells from protecting the body from other diseases, thus causing immunological failure. The most effective treatment currently in use is antiretroviral therapy (ART) (commonly known as “cocktail therapy”), which involves prescribing to the patient a combination of drugs that have a high likelihood of inhibiting HIV replication.
We first used source data of strains and their subsequent mutations from the Stanford HIV database.
Then iterated through each strain through simulated mutations (insertion, deletions, substitutions) to generate all possible strains in the next time step, then converged on the most likeley strains, initially through frequency count, then eventually using a reinforcement learning based approach based on historical mutation data (we called this method of converging on the strains as “Genetic Algorithmic Regressor”).