Genomic Surveillance of COVID-19 Variants With Language Models and Machine Learning

Abstract

The global efforts to control COVID-19 are threatened by the rapid emergence of novel SARS-CoV-2 variants that may display undesirable characteristics such as immune escape, increased transmissibility or pathogenicity. Early prediction for emergence of new strains with these features is critical for pandemic preparedness. We present Strainflow, a supervised and causally predictive model using unsupervised latent space features of SARS-CoV-2 genome sequences. Strainflow was trained and validated on 0.9 million sequences for the period December, 2019 to June, 2021 and the frozen model was prospectively validated from July, 2021 to December, 2021. Strainflow captured the rise in cases 2 months ahead of the Delta and Omicron surges in most countries including the prediction of a surge in India as early as beginning of November, 2021. Entropy analysis of Strainflow unsupervised embeddings clearly reveals the explore-exploit cycles in genomic feature-space, thus adding interpretability to the deep learning based model. We also conducted codon-level analysis of our model for interpretability and biological validity of our unsupervised features. Strainflow application is openly available as an interactive web-application for prospective genomic surveillance of COVID-19 across the globe.

Genetic justification of COVID‐19 patient outcomes using DERGA, a novel data ensemble refinement greedy algorithm

“…These models have incorporated various data sources, including comorbid diseases, 57,58 clinical factors, 59,60 genetic factors 39,42,61 and SARS-COV-2 viral clades. [62][63][64] Given the promising results obtained from therapeutic approaches, including complement inhibition, 65 in the treatment of COVID-19, the development of reliable prediction tools based on complement-related variants is of utmost importance.…”

The Epidemiology of Infectious Diseases Meets AI: A Match Made in Heaven

“…AI and machine learning can be used to analyze large datasets, such as genomic data, to identify patterns and trends relevant to the understanding and treatment of infectious diseases [ 9 , 10 , 11 , 12 ]. For example, machine learning algorithms have been utilized to identify potential drug targets for SARS-CoV-2, which causes COVID-19 [ 13 , 14 ].…”

Factors Modulating COVID-19: A Mechanistic Understanding Based on the Adverse Outcome Pathway Framework

“…For example, a recent study [ 164 ] highlights that through genomic surveillance it is possible to trace co-infections by distinct SARS-CoV-2 genotypes, which are expected to have a different impact on factors modulating COVID-19. Genomic surveillance of SARS-CoV-2 is able to reveal tremendous genomic diversity [ 165 ], and coupled with language models and machine learning approaches, contributes to predicting the impact of mutations (such as those occurring in the spike protein), and thus can better address challenging aspects, like an estimation of the efficacy of therapeutic treatments [ 166 ].…”