NEW YORK (GenomeWeb) – A team at the University of Pittsburgh has evaluated the Oxford Nanopore MinIon sequencer for characterizing near full-length 16S rRNA PCR amplicons from microbiomes.
The researchers, led by Kyle Bibby of the department of civil and environmental engineering, published their results yesterday on the BioRxiv preprint server.
"These results highlight the great potential of nanopore sequencing to analyze broad microbial community trends, and the challenge of applying nanopore sequencing to discern rare taxa in mixed microbial communities," they wrote.
For their evaluation, the researchers sequenced eight 16S rRNA libraries on the MinIon but only analyzed data from three because of a sample carryover problem. The libraries came from two pure culture samples, of Escherichia coli and Pseudomonas fluorescens, and a low-diversity sample of water from hydraulic fracturing.
They first evaluated how operational taxonomic unit (OTU) picking methods could be applied to nanopore data and found that OTU picking in conjunction with a reference library failed, while de novo OTU clustering assigned all sequences from the samples to unique OTUs. "These results highlighted the challenge of clustering reads from long, error-prone sequences, and necessitated analyzing the taxonomy of sequences individually without OTU picking," they wrote.
Next, they evaluated whether they could accurately annotate the taxonomy of nanopore 16S rRNA sequences from pure culture using three different annotation approaches. With the best performing one, the annotation accuracy was 96.7 percent at the phylum and 81.9 percent at the genus level for E. coli, and 90.9 percent at the phylum and 82 percent at the genus level for P. fluorescens.
Finally, using the environmental water sample from hydraulic fracturing, they compared nanopore and Illumina sequencing. The nanopore platform identified five phyla, while the Illumina platform identified 11. Both platforms found four phyla that together accounted for more than 99 percent of relative sequence abundance, and they both uncovered a similar community structure in the water sample. Phyla that only one platform detected accounted for less than 0.5 percent of relative sequence abundance.
Further comparisons of the data revealed that both platforms produced reproducible taxonomic assignments.
In addition, the group found that more than 10 percent of nanopore sequence reads came from carryover between runs.
Future developments should include strategies to exclude carryover between runs, improved bioinformatics to exclude poor-quality sequences, and a 16S rRNA annotation pipeline that provides an output in near real-time, the authors suggested. "Ultimately, the development of a rapid and low-cost microbiome approach will facilitate the application of clinical and environmental microbiome technologies," they wrote.