Rapid prediction of SARS-CoV-2 variant fitness using early frequency data

Introduction: The emergence of SARS-CoV-2 Variants of Concern (VOCs) has provided a powerful empirical demonstration of the importance of a phenotypic perspective in evolutionary dynamics. Materials and methods: This study utilizes high-resolution genomic surveillance data from the USA, UK, Germany, and Denmark (extracted from the CoVariants database) to model the competition between SARS-CoV-2 variants. The analytical approach bridges classical population genetics—specifically the Price equation and Fisher’s Fundamental Theorem—with epidemiological growth frameworks. A recursive fitting procedure was applied to the initial empirical frequency data of emerging Variants of Concern (VOCs) to assess the reliability of early-stage fitness estimates compared to full-series benchmarks. Results: The analysis demonstrates that fitness estimates derived from as few as the first 3 to 5 weeks of genomic data show a rapid convergence toward final values. These early estimates remain highly stable across diverse geographic regions, suggesting that the intrinsic transmission advantage of a VOC is a dominant driver. The results indicate that the selective coefficient of a variant remains largely decoupled from local environmental noise and stochastic fluctuations once a critical frequency threshold is surpassed. Conclusions: This study shows that the spread advantage of new COVID-19 variants is stable and can be accurately predicted using only early infection data, bypassing the need for long-term tracking. By analyzing genomic data from four countries, this method provides a crucial early warning window to help public health officials prepare for new variants before they become dominant.