Ultra-deep sequencing reveals intra-host diversity and selective constraint in SARS-CoV-2 samples from Chile

SARS-CoV-2 genomic surveillance has mainly focused on consensus genomes, but low-frequency variants within infected individuals can provide additional insight into viral diversity, mixed infections, and early evolutionary processes. Here, we performed ultra-deep whole-genome sequencing of 96 SARS-CoV-2 samples collected in Chile between 2020 and 2022 using the MGI ATOPlex platform and DNBSeq-G400 sequencing. The dataset reached an average depth of approximately 60,000×, enabling conservative detection of intra-host single-nucleotide variants (iSNVs). After stringent filtering, we identified 2,890 primary iSNVs. These variants were mostly present at low-to-intermediate allele frequencies, were well supported by sequencing depth, and showed a mutation spectrum dominated by C > T changes, consistent with previous SARS-CoV-2 within-host studies. Using a lineage-specific mutation-matching approach, we detected 13 candidate co-infections, supported by minor-lineage mutations at subconsensus frequencies. These events were compatible with periods of lineage overlap and replacement in Chile. Bayesian modeling suggested modest effects of Ct and sequencing depth on iSNV burden, while Omicron and candidate co-infection samples showed trends toward higher iSNV counts. Finally, iSNV-based dN/dS analysis showed that intra-host SARS-CoV-2 evolution was dominated by purifying selection, with localized nonsynonymous enrichment in Spike and selected accessory genes. Candidate co-infections showed higher apparent nonsynonymous signal in several loci, although this may partly reflect lineage mixing. Together, our results provide a high-resolution view of SARS-CoV-2 intra-host diversity in Chile and highlight the value of ultra-deep sequencing as a complement to consensus-level genomic surveillance.