Metagenomic Nanopore sequencing of influenza virus direct from clinical respiratory samples

Kuiama Lewandowski; Yifei Xu; Steven T. Pullan; Sheila F. Lumley; Dona Foster; Nicholas Sanderson; Alison Vaughan; Marcus Morgan; Nicole Bright; James Kavanagh; Richard Vipond; Miles Carroll; Anthony C. Marriott; Karen E Gooch; Monique Andersson; Katie Jeffery; Timothy EA Peto; Derrick W. Crook; A Sarah Walker; Philippa C. Matthews

For the Hazara virus control (10 4 genome copies/ml spike), genome coverage was 81.4-99.4% 144 (at 1x depth) for pools 1 and 2. Coverage in the high-background pool 3 was more variable 145 (21.5-96.5%; Fig 3A) . Influenza A genome coverage at 10 6 copies/ml was ≥99.3% for each 146 segment in all samples (Fig 3A) . At 10 4 genome copies/ml of influenza, mean 1x coverage per 147 segment was 90.3% for pools 1 and 2, but was substantially reduced in the high-background 148 pool 3 to 5.7% (Fig 3A) . At influenza titres <10 4 copies/ ml, coverage was highly variable across 149 . CC-BY-NC 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/676155 doi: bioRxiv preprint genome segments. However, when present at 10 3 copies/ml, 2/3 pools had sufficient data for 150 correct subtyping as H3N2 (Table 1) total reads generated was 1.0x10 5 to 2.4x10 6 (median 3.8x10 5 ) (Table S1) . GeneXpert-negative samples (Table S1 ). Based on this small dataset, sensitivity is 83% and 167 specificity 100% (95% CI 67-93% and 69-100%, respectively). There was a strong correlation 168 between Ct value and both reads per sample classified as influenza (R-squared =0.604) and 169 number of influenza reads per million reads (R-squared =0.623) (Fig 4) . 170 171