Limited value of Nanopore adaptive sampling in a long-read metagenomic profiling workflow of clinical sputum samples.

BACKGROUND: Oxford Nanopore adaptive sampling (NAS) is a method by which the long-read sequencing flowcell accepts or rejects DNA molecules that are actively being sequenced based on their initial ~ 500 bp sequences, selectively increasing target data output. NAS promises up to 5-10 × enrichment of target sequencing yield without additional sample preparation, but this optimal performance is dependent on ideal sample parameters which may be difficult to achieve under many real-world use-cases. We evaluated the use of NAS for profiling clinical sputum metagenomes. METHODS: We sequenced DNA extracted from clinical sputa and spike-in controls of a mock community of bacterial respiratory pathogens, using the current R10.4.1 MinION flowcell chemistry. RESULTS: We achieved at best 3.1 × enrichment of bacterial sequence output with NAS due to the shorter read lengths (~ 2.5 kb) from the PCR amplification necessary to compensate for low DNA extraction yields. More critically, we encountered rapid pore loss during our runs that reduced total sequencing yield by an estimated 80%. We were unable to mitigate the pore loss despite extensive attempts to reduce contaminant carry-over, and we could not determine its cause but ruled out NAS and pore underloading as contributing factors. CONCLUSIONS: We conclude that the utility of NAS is often limited by the characteristics of the metagenomic sample studied, and that the factors contributing to pore loss need to be resolved before ONT sequencing can be reliably applied to long-read metagenomics.