Unbiased metagenomic detection of RNA viruses for rapid identification of viral pathogens in clinical samples (preprint)

Unbiased long read sequencing approaches for clinical metagenomic sample analysis holds enormous potential for pathogen detection, including improved detection of unknown, novel or emerging viruses. However, the rapid rate of development in nanopore sequencing and library preparation methods complicates the process of selecting a standardized method for unbiased RNA virus detection. Here, we evaluate multiple sequencing approaches to identify a workflow with sufficient sensitivity, limits of detection, and throughput for potential utilization in a clinical laboratory setting. Four separate library preparation methods for the Oxford Nanopore Technologies MinION sequencer are compared, including direct RNA, direct cDNA, rapid cDNA, and double stranded cDNA. We also establish that depletion of host RNA is not required and can be deleterious for viral RNA detection in some instances when using samples in viral transport media (VTM) or plasma. Using unbiased whole genome amplification following reverse transcription, we achieve limits of detection on the order of 1.95E03 GE/mL of Venezuelan Equine Encephalitis Virus (VEEV) spiked in human plasma. We also report initial detection of 5.43E06 GE/mL of coronavirus 229E spiked into VTM samples containing human background RNA which are expected to decrease significantly during upcoming testing. These metrics were achieved within a 6-plex multiplex reaction, illustrating the potential to increase throughput and decrease costs for relevant sample analysis. Data analysis was performed using EPI2ME Labs framework and open access tools that are readily accessible to most clinical laboratories. Taken together, this work describes an optimized method for unbiased nanopore sequencing and analysis of RNA viruses present in two common clinical matrices.

Validation of an unbiased metagenomic detection assay for RNA viruses in viral transport media and plasma (preprint)

Unbiased long read sequencing holds enormous potential for the detection of pathogen sequences in clinical samples. However, the untargeted nature of these methods precludes conventional PCR approaches, and the metagenomic content of each sample increases the challenge of bioinformatic analysis. Here, we evaluate a previously described novel workflow for unbiased RNA virus sequence identification in a series of contrived and real-world samples. The novel multiplex library preparation workflow was developed for the Oxford Nanopore Technologies (ONT) MinION sequencer using reverse transcription, whole genome amplification, and ONT's Ligation Sequencing Kit with Native Barcode Expansion. The workflow includes spiked MS2 Phage as an internal positive control and generates an 8-plex library with 6 samples, a negative control and a gfp transcript positive control. Targeted and untargeted data analysis was performed using the EPI2ME Labs framework and open access tools that are readily accessible to most clinical laboratories. Contrived samples composed of common respiratory pathogens (Influenza A, Respiratory Syncytial Virus and Human Coronavirus 229E) in viral transport media (VTM) and bloodborne pathogens (Zika Virus, Hepatitis A Virus, Yellow Fever Virus and Chikungunya Virus) in human plasma were used to establish the limits of detection for this assay. We also evaluated the diagnostic accuracy of the assay using remnant clinical samples and found that it showed 100% specificity and 62.9% clinical sensitivity. More studies are needed to further evaluate pathogen detection and better position thresholds for detection and non-detection in various clinical sample metagenomic mixtures.