Tonight, I watched a Nanopore webinar by Tianyuan Zhang. The session’s title is “The Newest Oxford Nanopore R10.4.1 Full-length 16S rRNA Sequencing Enables the Accurate Resolution of Species-Level Microbial Community Profiling.” Zhang spoke about the resolution gains with full-length 16S sequencing, citing a Nature article from 2019. However, the error rate of ONT was previously a challenge. The objective of Zhang’s study was to test the accuracy of R10.4.1 flow cells with Q20+ chemistry in terms of performance, error distribution and accuracy, detection at the species and genus levels, and a comparison of PacBio and ONT workflows. The team made a synthetic community with different genomic DNAs and a range of differences in copy number and abundance. The error profiles were compared, and Zhang mentioned that the mismatch error rate was reduced to less than half (R10.4.1 vs 9.4.1), the insertion error rate was reduced to one-third. The proportion of correctly classified reads depended in part on the database used: NCBI and SILVA. PacBio was better than Nanopore when it comes to recall. PacBio “gets far fewer false positives than Nanopore” based on one of the tests Zhang and team conducted. Zhang concluded that the NCBI database was best for species-level detection with long reads. Zhang concluded that for environmental samples, R10.4.1 obtained more species than PacBio. Both PacBio and R10.4.1 produced similar Shannon and Simpson indices. Among the major species, R10.4.1 and PacBio platforms revealed similar microbiomes. Zhang also noted that both classifiers and databases used did affect species composition.
