Comparison of nanopore methods for respiratory clinical metagenomics

Respiratory metagenomic sequencing is increasingly used for rapid, pan-microbial diagnosis in clinical practice, but there is a need to evaluate different wet lab protocols for performance and feasibility. We compared the Oxford Nanopore Technologies dual-arm protocol developed for surveillance, which processes paired cellular and acellular fractions, with the clinically deployed pan-microbial mechanical lysis protocol. Thirty-seven respiratory specimens were processed in parallel. Time from processing start to start of sequencing was 8.5 h for the dual-arm workflow versus 6 h for the pan-microbial protocol used in clinical practice. Sample failures, defined as internal control non-detection in an otherwise negative sample, occurred in 0/37 (0%) samples for the pan-microbial protocol, compared with 1/37 (3%) in the dual-arm acellular arm and 8/37 (22%) in the cellular arm. In evaluable comparisons against routine laboratory results, overall percent agreement with the dual-arm workflow was 95% for bacteria (95% CI 75 – 99), 79% for fungi (60 – 91) and 75% for viruses (47 – 91), compared to 88% (71 – 96), 88% (71 – 97) and 77% (50 –92) with the pan-microbial protocol. Genome recovery at 1x and 10x depth was not significantly different between the workflows. Overall, the dual-arm workflow demonstrated longer turnaround time and higher failure rates, with a similar performance in successful samples, indicating that further optimisation of the dual-arm protocol is required for translation into a clinical service setting.