Skip to main content
Premium Trial:

Request an Annual Quote

Following Publication, UW Team Pursues Better Accuracy for De Novo Nanopore Sequencing


NEW YORK (GenomeWeb) − Following their recent publication in Nature Biotechnology that demonstrated MspA nanopore sequencing of bacteriophase phi X 174 DNA and proof-of-concept for hybrid assembly and SNP detection, researchers at the University of Washington are improving the technology for de novo sequencing.

The team, from the groups of Jens Gundlach in the department of physics and Jay Shendure in the department of genome sciences at UW, published an almost identical version of the paper on the arXiv e-print repository last month. Because of their pending journal publication, the scientists were not able to comment on their work at the time.

For their study, they generated reads up to 4,500 bases in length by feeding adaptor-modified DNA, controlled by phi29 DNA polymerase, through the MspA pore and measuring changes in the ion current.

They found that the current is determined by four nucleotides at a time, with each of the 256 possible quadromers generating a unique current value. After generating a quadromer map from known sequences, they analyzed the 10.7-kilobase phi X 174 genome, generating more than 100 long reads, 10 percent of them longer than 3,000 bases, and the longest about 4,500 bases in size.

They also generated a hybrid assembly of the phi X 174 genome, using a single long nanopore and short Illumina reads. Further, they found that a short region from three nanopore reads aligned to the phi X 174 genome with high confidence, picking it out from a database of more than 5,000 viral genomes. In addition, they found that they could detect about three quarters of artificially introduced SNPs using nanopore data.

However, the calling accuracy was not good enough yet for de novo sequencing, and the focus of their current research is to enable this, Gundlach told In Sequence. "We have two tricks up our sleeves that are novel, but at this point I can't quite divulge how it works," he said.

One error source that limits the calling accuracy, he said, is the fact that the polymerase, which feeds the DNA through the pore in a stepwise fashion, sometimes makes erratic steps forward or backward, so that a base is missed or the same base is read several times in a row. "Those kinds of errors are at present limiting things the most, so that's what we are working on with other enzymes," Gundlach said.

In the paper, he and his colleagues suggested that using a helicase instead of a polymerase might improve matters, and Gundlach said that this is "one of the next things" his team is pursuing. "But we have also found another trick that goes around this and may potentially enable de novo sequencing with high accuracy," he said.

Another problem is that the polymerase sometimes falls off, limiting the read length. Since they submitted their paper, the researchers have been able to achieve "substantially longer" reads than the 4,500-base read they reported, he said, but he declined to say what the maximum read length is.

For their published and current work, Gundlach, a physicist by training, teamed up with Jay Shendure, a colleague at UW, whose group provides expertise in biochemistry and genomics. "It's a really nice and fruitful collaboration," he said.

Their research is unrelated to work by Illumina to develop a commercial nanopore sequencer. Last year, Illumina exclusively licensed intellectual property related to MspA-based nanopore sequencing developed by Gundlach and his colleague Michael Niederweis from the University of Washington and the University of Alabama.

The published work "was entirely done at the University of Washington, with technology that we have developed in the lab" and "has no overlap with Illumina," Gundlach said.

In addition, he said, it is unrelated to Oxford Nanopore Technologies' research. "It was all done on our own and totally independent of what Oxford has done."

Oxford Nanopore and its collaborators and early-access customers have not yet published any nanopore sequencing data in a peer-reviewed journal. Earlier this year, David Jaffe, a collaborator from the Broad Institute, showed data from Oxford's MinIon platform at a conference, citing reads up to 15 kilobases in length, and early-access users of the technology recently started posting comments on the technology and data online, including an 8.5-kilobase read. In addition, in a conference presentation this year, Chief Technology Officer Clive Brown said the firm had obtained reads in excess of 50 kilobases.

Commenting on the UW publication in a Twitter post last month, Brown said that "most of this work [was] previously disclosed, [and has] not [been] properly accredited in this paper" and suggested the UW group might have engaged in "plagiarism." A spokesperson for Oxford Nanopore declined to comment, adding that Brown had expressed his personal views, not those of the company.

Gundlach said the current traces of the data generated by his group and by Oxford Nanopore "look surprisingly similar."

One of the main differences between the technologies, he said, is that Oxford Nanopore has developed a highly parallelized platform with thousands of nanopores, whereas his group has been working with a single-channel system and a sole pore. Oxford Nanopore has not disclosed yet what types of nanopores or processive enzymes it is using.

Any commercial nanopore sequencer will be most useful for applications that require long reads, even if they have low accuracy, Gundlach said. "At present, the calling accuracy is not very impressive, but the read length is impressive."

He believes the technology will not lend itself to "brute-force" whole-genome sequencing but rather to targeted sequencing applications.

A major promise is the ability of nanopores to detect epigenetic modifications, he said, a capability his lab and another group demonstrated in separate publications last year. Pacific Biosciences' platform can also detect certain epigenetic modifications, but from a secondary measurement – the duration of the optical signal – rather than the primary signal, the dye color. Nanopores, on the other hand, measure epigenetic modifications directly from the current, the primary measurement, Gundlach said. "We believe that is an advantage of nanopore sequencing."