Direct comparison of clear DX Nanopore and Illumina sequencing of SARS-CoV-2

There are limited data directly comparing SARS-CoV-2 sequencing methods using two major commercial approaches, Oxford Nanopore Technologies Clear DX (ONTDX) and Illumina. RNA was extracted from 1,513 SARS-CoV-2 RNA-positive respiratory samples, split into two aliquots, and sequenced using both ONTDX and Illumina sequencing methods. FASTQ sequences generated by the ONTDX and Illumina (550 or 2,000) platforms used to test these samples were analyzed using either the same assembly and alignment strategy or using the platform default methodologies. ONTDX and Illumina sequencing results were not significantly different in generating consensus genomes, though slightly better coverage was obtained with ONTDX compared to Illumina. The results were similar using the harmonized assembly and alignment strategy and using the "real-world" platform default assembly and alignment. Approximately 2.5% of samples sequenced were assigned to different sublineages by the different methods, and two samples were assigned to different clades. Median sample storage time ranged from 594 to 970 days before RNA extraction and sequencing, yet >90% genome coverage was achieved in >91.5% of samples. ONTDX and Illumina-based deep sequencing platforms both generate high coverage sequences for SARS-CoV-2. ONTDX performed slightly better than Illumina-based sequencing. Despite a high degree of overall agreement, lineage and clade assignment may differ based on the sequencing and alignment methods used, illustrating the need for caution when comparing phylogenetic relationships based on sequences generated by different platforms. Respiratory samples were also stable and provided high-quality sequencing results following storage for up to 970 days.IMPORTANCESARS-CoV-2-positive respiratory samples from 1,513 individuals were sequenced using the Oxford Nanopore Clear DX platform. Matching aliquots for these were sequenced using either Illumina NextSeq 550 or NextSeq 2000. A harmonized bioinformatic approach compared raw FASTQ files generated by each platform. Differences in coverage and, surprisingly, phylogenetic lineage assignments were observed; thus, assembly and alignments were also compared using the platform default methods. Both platforms generated high coverage for SARS-CoV-2; the Nanopore method was as good, if not slightly better, than Illumina-based sequencing in these studies. Two of 1,546 samples were assigned to different clades by different platforms, and 2.5% of sequences were assigned to different sub-lineages when sequenced by different platforms. These data illustrate that different methods may lead to small differences in phylogenetic assignments and highlight one limitation in the interpretation of outbreak diversity.