A rapid, cost-effective tailed amplicon method for sequencing SARS-CoV-2

Abstract

Background: The global COVID-19 pandemic has led to an urgent need for scalable methods for clinical diagnostics and viral tracking. Next generation sequencing technologies have enabled large-scale genomic surveillance of SARS-CoV-2 as thousands of isolates are being sequenced around the world and deposited in public data repositories. A number of methods using both short- and long-read technologies are currently being applied for SARS-CoV-2 sequencing, including amplicon approaches, metagenomic methods, and sequence capture or enrichment methods. Given the small genome size, the ability to sequence SARS-CoV-2 at scale is limited by the cost and labor associated with making sequencing libraries.

Results: Here we describe a low-cost, streamlined, all amplicon-based method for sequencing SARS-CoV-2, which bypasses costly and time-consuming library preparation steps. We benchmark this tailed amplicon method against both the ARTIC amplicon protocol and sequence capture approaches and show that an optimized tailed amplicon approach achieves comparable amplicon balance, coverage metrics, and variant calls to the ARTIC v3 approach.

Conclusions: The tailed amplicon method we describe represents a cost-effective and highly scalable method for SARS-CoV-2 sequencing.

Keywords: COVID-19; Genome sequencing; SARS-CoV-2; Viral surveillance.

Fig. 1

Methods for SARS-CoV-2 genome sequencing compared in this study. a In Illumina’s Nextera DNA Flex Enrichment protocol cDNA is tagmented and made into barcoded sequencing libraries, which are then enriched using sequence capture with a respiratory virus panel containing probes against SARS-CoV-2. b In the ARTIC protocol, first strand cDNA is enriched by amplifying with two pools of primers to generate amplicons tiling the SARS-CoV-2 genome. These amplicons are then subjected to either Illumina or Oxford Nanopore library preparation, using methods that either directly add adapters to the ends of the amplicons or to fragment them to enable sequencing on a wider variety of Illumina instruments. c The tailed amplicon approach, developed here, enriches first strand cDNA using ARTIC v3 primers containing adapter tails. This allows functional sequencing libraries to be created through a second indexing PCR reaction that adds sample-specific barcodes and flow cell adapters

Fig. 2

Comparison of sequence capture, ARTIC v3 amplicon, and tailed amplicon workflows on SARS-CoV-2 isolate. a Percentage of the BEI WA1 isolate genome coverage at 10x at different subsampled read depths when sequenced with the indicated approach. b Percent of the BEI WA1 isolate genome coverage at 100x at different subsampled read depths when sequenced with the indicated approach. c Observed read depth for each of the expected amplicons for the BEI WA1 isolate amplified with the ARTIC v3 protocol at a subsampled read depth of 100,000 raw reads. d Observed read depth for each of the expected amplicons for the BEI WA1 isolate amplified with the tailed amplicon v1 (2 pool amplification) protocol at a subsampled read depth of 100,000 raw reads. e Observed read depth for each of the expected amplicons for the BEI WA1 isolate amplified with the tailed amplicon v2 protocol (4 pool amplification) at a subsampled read depth of 100,000 raw reads. f Variants detected for the BEI WA1 isolate at a read depth of up to 1,000,000 raw reads (or the maximum read depth for the sample if below 1,000,000 reads). The positions of all variants detected in this study are shown and bases where the sample matches the Wuhan-Hu-1 reference are shown in grey

Fig. 3

Comparison of sequence capture, ARTIC v3 amplicon, and tailed amplicon workflows on clinical specimens spanning a range of viral loads. a Samples with N1 and N2 Ct values ranging from approximately 20–35 chosen for testing of SARS-CoV-2 sequencing workflows. Samples are colored as in panels c-f. b Evenness of representation of amplicons for different workflows as a function of sample N1 Ct value. Percentage of genome coverage at 100x at different subsampled read depths for each sample when sequenced using the following approaches: c Illumina Nextera DNA Enrichment; d ARTIC v3 with TruSeq library preparation. e Tailed amplicon v1 (2 pool amplification); f Tailed amplicon v2 (4 pool amplification)

Fig. 4

Variants detected using different sequencing workflows. Variants detected for the indicated sample and sequencing protocol at a read depth of up to 1,000,000 raw reads (or the maximum read depth for the sample if below 1,000,000 reads). The positions of all variants detected in this study are shown and bases where the sample matches the Wuhan-Hu-1 reference shown in grey