RAPID prep: A Simple, Fast Protocol for RNA Metagenomic Sequencing of Clinical Samples

Abstract

Emerging infectious disease threats require rapid response tools to inform diagnostics, treatment, and outbreak control. RNA-based metagenomics offers this; however, most approaches are time-consuming and laborious. Here, we present a simple and fast protocol, the RAPIDprep assay, with the aim of providing a cause-agnostic laboratory diagnosis of infection within 24 h of sample collection by sequencing ribosomal RNA-depleted total RNA. The method is based on the synthesis and amplification of double-stranded cDNA followed by short-read sequencing, with minimal handling and clean-up steps to improve processing time. The approach was optimized and applied to a range of clinical respiratory samples to demonstrate diagnostic and quantitative performance. Our results showed robust depletion of both human and microbial rRNA, and library amplification across different sample types, qualities, and extraction kits using a single workflow without input nucleic-acid quantification or quality assessment. Furthermore, we demonstrated the genomic yield of both known and undiagnosed pathogens with complete genomes recovered in most cases to inform molecular epidemiological investigations and vaccine design. The RAPIDprep assay is a simple and effective tool, and representative of an important shift toward the integration of modern genomic techniques with infectious disease investigations.

Keywords: RNA sequencing; diagnostics; infectious diseases; metagenomics.

Figure 1

RAPIDprep development experiments. All results here are derived from the same sample extracts (RESP01-RESP03) run in duplicate and presented as mean values and error as standard deviation (SD). (A) The shaded bars are representative of the percentage of residual rRNA reads in the library following rRNA depletion with either an in-reaction cDNA synthesis method (grey) or a pre-cDNA hybridization approach (orange). The bars are clustered with respect to the sample they are derived from, labeled on the X-axis. (B) A comparison in total library yield, in nanomolar generated using Tapestation values, following a parallel experiment with a one-step and two-step second strand synthesis step using the Sequenase enzyme. The grey and orange shaded bars are representative of the one-step and two-step protocols, respectively. (C) Grey-shaded bars represent the total library yield of each sample under different library amplification cycling conditions. The X-axis is marked with the number of amplification cycles and is sub-grouped by source sample. (D) The duplication rate of reads generated in the final libraries following cycle titration; the number of cycles for each sample is indicated on the X-axis, and is sub-grouped by source sample.

Figure 2

Filtered read distribution and classification across forty RAPIDprep libraries. The sequence reads were classified into five categories: low-quality reads (blue), human rRNA reads (red), human non-rRNA (pink), non-human rRNA reads (green), and non-human non-rRNA reads (light green). Low-quality, human rRNA, human non-rRNA, and non-human rRNA were excluded from downstream analysis, and the non-human non-rRNA reads were the sole target reads for pathogen detection. Relative distribution was calculated by dividing the number of reads mapping to the relative category by the total number of reads for the individual library, before conversion into a percentage by multiplying the value by 100. The results were ordered by library number and grouped by sample type with a further key in grey shaded indicating the sample extraction platform used.

Figure 3

Quantitative detection of SARS-COV-2 and RSV sequences. A simple linear-regression model was applied to both SARS-CoV-2 (A) and RSV (B) data sets with a line of best fit estimating the relationship between log-transformed reads per million (_logRPM) and cycle threshold (CT) values. The linear-regression slope coefficient and the intercept parameter are printed on the top right of each plot, with R² calculated to measure the goodness of fit.

Figure 4

Comparison of RAPIDprep to commercial RNA library preparation kit. Using previously generated data for the kids SARI cohort, we compared the 24 most abundant species identified across both protocols for the same set of samples. An unclustered heatmap of microbial abundance (Z-score) is shown, with differences between samples identified by a deeper blue shading, while organisms conserved across samples are lighter blue through to red. A frequency histogram is overlayed on the color key and signifies the count of each Z score at any given point. Tick labels on the X-axis in the ICUXX format represent deep-RNA sequencing generated previously, while tick labels in the RAPIDXX format represent sequencing data generated in this study using the RAPIDprep assay for the corresponding samples.