Using time-series genomic data for early detection of novel SARS-CoV-2 variants with increased transmissibility

Rapidly identifying variants of a pathogen that spread more quickly is an important public health challenge that could inform outbreak control policies and vaccine design. Genetic information, such as the presence of repeat mutations observed in multiple viral lineages, could help to distinguish between novel variants with increased transmission and those that are unconcerning. However, most current methods for detecting highly transmissible variants do not account for genetic similarities or differences between variants, often using only variant frequencies over time to infer transmission advantages. Here we develop an epidemiological model that describes the evolution of a pathogen population over time. This model accounts for the effects of individual mutations on disease transmission. From this model, we derive a method for inferring the fitness effects of different mutations from time-series genomic data. Applying this method to SARS-CoV-2 time-series sequence data, we detect variants that are likely to increase transmission while they are still at low frequencies, even 1-2%.

The method developed is general and could be applied to study the fitness effects of mutations in other contexts.