Machine Learning-Based Prediction Of Viral Mutations From Genomic Sequences

SUMMARY

• There have been significant research efforts to develop machine learning and predictive analytics based tools to predict viral evolution which remains the main obstacle in the early detection of drug-resistant strains and facilitate the design of more efficient antiviral treatments.
• Machine learning and advanced algorithms have facilitated the study of structure-activity relationships, secondary and tertiary structure evolution prediction, and sequence error correction.
• This technology includes a software tool for predicting the dominant circulating strain of evolving pathogens like influenza and help with vaccine design.
• The underlying algorithm in this invention reverse‐engineers the laws driving evolutionary changes in evolving pathogens to predict future mutations and emerging strains.

FIGURE

ADVANTAGES

ADVANTAGES

• First of its kind prediction approach for evolution of viruses

• Performs quantitative assessment indicative of the risk of viral emergence based on strains that circulate in the wild.   

• This tool predicts dominant strains of future seasonal epidemics significantly better than the WHO recommendations used today in flu shot compositions.

APPLICATIONS

• Prediction of dominant strains of future seasonal epidemics significantly better than the WHO recommendations used today in flu shot compositions.

• Vaccine development focused on new emerging strains of viruses.

• The software can also be used by health agencies like CDC to take preventive measures and issue guidelines based on new strains and potential impact on population. 

PUBLICATIONS

• Preparing For the Next Pandemic: Learning Wild Mutational Patterns At Scale For Analyzing Sequence Divergence In Novel Pathogens.  Jin Li, Timmy Li, Ishanu Chattopadhyay. medRxiv 2020.07.17.20156364; doi: https://doi.org/10.1101/2020.07.17.20156364

April 14, 2022

Proof of concept

Patent Pending

Licensing,Co-development

Ishanu Chattopadhyay