CLAE: A High-Fidelity Nanopore Sequencing Strategy for Read-Level Viral Variant Detection and Environmental RNA Virus Discovery.

High-fidelity (HF) long-read sequencing enables accurate profiling of microorganisms and pathogens at single-molecule resolution. However, current Oxford Nanopore Technologies (ONT)-a revolutionary platform offering real-time, portable sequencing at relatively low instrumental cost-suffer from severe read-length bias, limited accuracy (often <Q20), and low throughput. Here, Circular- and Linear-Amplicon-Mediated Error Correction (CLAE) is introduced, a biochemical and computational approach that addresses these limitations by integrating hairpin ligation, pre-circling, single-stranded DNA linearization, and targeted nickase-based debranching. CLAE significantly enhances rolling-circle amplification (RCA) efficiency for long DNA templates, markedly improving Nanopore sequencing yield and accuracy. CLAE achieves Q30-level accuracy in up to 27% of RCA reads, throughput exceeding 800 Mb per 100 pores, and an N50 of ≈15 Kb (bacterial genome). Moreover, its bidirectional subreads and high throughput substantially boost accuracy without compromising read length. CLAE is validated by resolving SARS-CoV-2 quasi-species from community wastewater and recovering novel, full-length RNA virus genomes from marine samples. CLAE enables precise variant detection in complex samples and corrects short-read misassemblies, significantly broadening ONT's utility in metaviromics, epidemiology, and environmental surveillance. Thus, CLAE establishes a versatile, field-compatible platform for high-fidelity viral genome sequencing in targeted and agnostic contexts.