Most SARS-CoV-2 genomes were produced with tiled amplicons. Software unaware of primer schemes and error modes, together with outdated primers as variants evolved, generated systematic errors and false phylogenetic structure in waves. Viridian was developed for rigorous amplicon-aware reconstruction and applied to all public data through June 2024, producing high-quality consensus sequences and a global tree of 4,471,579 samples, supported by simulation and empirical validation and openly viewable online.
Key findings
- Primer mismatch and amplicon-unaware pipelines created variant-linked errors with global tree effects. Reassembly reduced systematic artifacts and yielded a higher-quality open phylogenetic resource.
Why this matters globally
Phylogenies underpin variant, transmission and mutation tracking. Correcting artifacts improves retrospective COVID-19 inference and offers a design pattern for genomic surveillance in future outbreaks.
Thai researcher contribution
Elizabeth M. Batty, Wanwisa Dejnirattisai and Juthathip Mongkolsapaya are affiliated with MORU and Siriraj Hospital, linking Thailand-based networks to global genomic infrastructure.
Limitations to consider
Public data inherit unequal sampling and raw-read deposition. Viridian reduces amplicon errors but cannot recover missing data or correct all metadata. Consensus sequences lose within-host variation, and processing 4.47 million samples requires substantial computation.