NEW YORK (GenomeWeb) – Researchers from the US Army's Edgewood Chemical Biological Center and Signature Science have shown that a six-hour amplicon sequencing run on Oxford Nanopore Technologies' MinIon can identify bacteria and viral strains.
The study, published this week in GigaScience, is the first to show the feasibility of amplicon sequencing on the MinIon, adding to a growing list of potential applications for the instrument.
Thus far, researchers have published or presented data on the instrument for de novo assembly of yeast and bacteria genomes, sequencing clinically relevant genes, surveying microbes in air samples, and more.
In addition, the new study shows that a targeted approach may help to get around the instrument's current high error rate to identify bacterial and viral species.
The researchers used 16S rDNA sequencing to identify an Escherichia coli strain down to its species level and amplicon sequencing to differentiate between three closely related poxviruses.
Sam Minot, a bioinformaticist at Signature Science, told GenomeWeb that the team decided to pursue an amplicon sequencing approach because it would enrich genomic regions with a "strong and characteristic signature, increasing the sensitivity of detection by reducing background clutter." In addition, amplicon sequencing may enable quicker pathogen identification than shotgun sequencing, particularly for complex mixed samples where the majority of the DNA is from the host or other organism, the authors wrote.
"More accurate identification and characterization can occur using the MinIon by concentrating the data collection over a particular genomic region via targeted amplification," they wrote.
The team amplified E. coli using 16S-specific primers and created poxvirus amplicons using consensus primers, which they used to amplify cowpox, vaccinia-MVA, and vaccinia-Lister strains.
They next used Oxford Nanopore's SQK-MAP-002 library prep kit and performed nanopore sequencing on the MinIon using the R7.0 flowcell.
For each sample, they generated between 296 and 1,335 reads with mean read lengths between 770 bp and 2,863 bp.
The researchers tested both BLASR, which was developed to work with Pacific Biosciences' long reads, and the LAST aligner, which was developed by the Computational Research Center in Japan, to align the reads to the reference genomes, finding that while both were comparable there were slight differences.
For instance, LAST aligned between 47 and 751 reads per sample, compared to between 18 and 270 reads with BLASR. However, for both aligners, a large proportion of aligned reads completely spanned the amplicon.
The authors found that although the error rate of the MinIon was about 30 percent, they were still able to identify the correct species because the number of reads that aligned to the correct strain and species was still much higher than the number of incorrect alignments.
When aligning reads to the reference database for the E. coli sample the team used all 16S rDNA, and for the viral reference database they included all poxvirus sequences that could be potentially amplified with the consensus primers. When aligning the reads, they found that for each sample the nanopore-generated reads identified the correct genus and species.
"Vaccinia-Lister and vaccinia-MVA are over 98 percent similar to each other, yet these were correctly differentiated based on the mapping characteristics of the datasets for each sample," the authors wrote.
Josh Quick, a doctoral researcher at the University of Birmingham in Nick Loman's lab — another early-access participant — said that using the MinIon to "classify organisms by aligning to a database of known sequences is likely to be a successful approach" since aligners like LAST are able to still make correct alignments even in reads with errors.
Quick said that his lab has also found that sequencing PCR-amplified DNA on the MinIon can help improve accuracy, however, the downside is that it can be "quite laborious and somewhat negates the benefit of having a real-time sequencer."
He added that the MinIon technology is improving rapidly. For instance, accuracy has already improved greatly since the Edgewood Chemical team's report. With the newest library prep kit, SQK-MAP-004, and most recent flowcell, R7.3, he said his lab has achieved reads with 90 percent or greater accuracy.
A number of other early-access users reported on their experiences with the instrument at the Advances in Genome Biology and Technology in February. Sara Goodwin from Cold Spring Harbor Laboratory reported there that flow cells, while still somewhat variable, have improved over the course of three iterations.
The Edgewood Chemical team, which used the second version flow cell, reported in the GigaScience study that the vast majority of nanopores on the flow cell were unusable — with only 50 to 100 active channels out of 512 total per run. Goodwin said in February that having at least 350 to 400 active pores results in good performance.
Minot agreed that the newer chemistry and flow cell would result in improved accuracy and throughput, and said that his group plans to keep working on improving amplicon sequencing on the MinIon with the ultimate goal of using it to provide point-of-care diagnostics.
The Genome Analysis Centre in the UK is also looking to use the MinIon in the field in a project to sequence microbes in the air. Richard Leggett, project leader in the data infrastructure and algorithms group at TGAC, told GenomeWeb that the institute has had a "long-running aim to literally put a sequencer in the field," a goal that "seems more realistic" with the development of a nanopore sequencer that is small and requires little power. However, he said that improvements to the sample prep process will be needed, particularly miniaturization, which could be accomplished by incorporating microfluidics.