Complementary Use of Shotgun Metagenomics and Multiplex PCR for Pathogen Detection on High-Touch Surfaces

Shotgun metagenomics is being increasingly promoted as a pathogen-agnostic alternative to targeted PCR for environmental surveillance, yet its practical value on dry, low-biomass, host-dominated surfaces remains uncertain. In a companion study, we reported QIAstat-Dx multiplex PCR screening of 55 pooled high-touch surface samples from Riyadh, identifying 19 positive pools (34.5%). Here, in a retrospective exploratory comparison, we use those previously reported PCR calls as the targeted comparator and ask what unenriched shotgun metagenomic sequencing adds. Ten archived pools, each with at least one PCR-detected viral target, were sequenced (Illumina NovaSeq X Plus, 2 × 150 bp). Human-mapped reads constituted a substantial fraction of every library (mean 59.1%; range 42.1–80.2%). Among 14 evaluable PCR-positive target events, metagenomics assigned at least one read pair to the corresponding target taxon in 10 (71.4%; exact 95% CI, 41.9–91.6%). Because the cohort was preselected for viral PCR positivity, this value represents positive agreement/taxonomic concordance within the selected cohort and not diagnostic sensitivity. At the same ≥1 read-pair threshold, target taxa were also assigned to 91 of 246 (37.0%) evaluable pool-by-target cells in which PCR was negative; raising the threshold to ≥10 read pairs reduced positive agreement to 35.7% while background assignment remained 24.8%. Of six target categories with PCR-positive events, only coronavirus 229E was assigned more read pairs in its PCR-positive pool than in any PCR-negative pool. Across all ten pools, the three respiratory RNA-virus taxa represented among the PCR-positive calls accounted for only 105 classified read pairs in total—including assignments in both PCR-positive and PCR-negative pool-target cells—equivalent to 0.12% of the viral fraction. Even the broadest viral reference coverage was limited, as adenovirus in HP10 reached 16.39% (23 distinct alignment spans), while SARS-CoV-2 in CP17 reached 8.83% (12 spans); these partial, discontinuous alignments did not support robust whole-genome consensus reconstruction or genome-wide variant or phylogenetic analysis. Metagenomics nevertheless provided a broader taxonomic context beyond the targeted panel. Under this retrospective unenriched workflow, target recovery was variable, no tested assignment threshold separated a PCR-positive target signal from the background assignment, and shotgun metagenomics added ecological context rather than providing a reliable replacement for targeted multiplex PCR in this dataset.