NEW YORK (GenomeWeb) – A new nanopore sequencing method called "Read Until" that was developed by researchers at the University of Nottingham in the UK could have applications in metagenomics, haplotyping, and eventually in infectious disease diagnostics, among other applications.
The researchers developed the method, which enables selective sequencing without using targeted capture methods, for Oxford Nanopore Technologies' MinIon device.
Matt Loose, an associate professor at the University of Nottingham and the senior author of the paper that described the method in Nature Methods today, first presented the approach at the Advances in Genome Biology and Technology meeting in February, and has received a three-year grant from the UK's Biotechnology and Biological Sciences Research Council to continue developing applications for the method.
The Read Until method essentially seeks to only sequence DNA from a genome of interest in a mixed sample, or from a specific region of a single genome, without having to do a targeted capture or enrichment.
To accomplish this, the researchers first convert the reference genome into artificial raw data from the MinIon, which they called "squiggle" data — the current trace generated from translocating DNA through the nanopore. Then, the software they developed can decide in real time whether a DNA amplicon that is being sequenced appears to match the desired DNA region or reference genome. If so, it will continue to sequence it, but if not, it will stop sequencing that piece of DNA and start sequencing a new one.
In the study, the researchers first demonstrated that they could selectively target two 5-kilobase regions of the lambda phage genome, sequencing until they achieved the desired coverage of those regions. In addition, they were able to tweak the software to switch to a different region once they reached their coverage goal of the initial two regions.
Loose said that the concept for Read Until was initially described by Oxford Nanopore, but had not been implemented by the company. Loose and his team were inspired to attempt to design the computational tool after learning that Josh Quick, a graduate student in Nick Loman's University of Birmingham laboratory, had taken the MinIon to Guinea to sequence samples of patients suspected to be infected with Ebola during the outbreak in West Africa last year.
The group was often doing the sequencing in remote areas, and reported that one of the biggest challenges was the availability of a stable Internet connection to do base and variant calling and to send data for further phylogenetic analysis.
"They couldn't look and see what they were sequencing in real time," Loose said.
As such, Loose and his group wanted to develop a protocol that could work offline. The "squiggle matching approach does not rely on base calling in a cloud," Loose said. Even without an Internet connection, a sequence experiment can be designed to target the Ebola genome from a mixed sample with a high background of human DNA, so could at the very least provide a yes or no answer to whether a sample was positive for Ebola.
Now, said Loose, researchers are using the method in Brazil, where they are sequencing samples to look for the Zika virus.
"We're just scratching the surface of what we might be able to do with something like this," Loose said.
The next step is to transition from matching squiggles to doing real-time base calling as the DNA molecule runs through the pore.
Oxford Nanopore has taken the first step in this direction with the early-access launch of a local basecalling tool through MinKnow, which does not require the cloud. Loose said that his lab plans to test the local basecaller and try to make it compatible with the Read Until feature. "We haven't done the next step of basecalling as the molecule runs through the pore," but "it won't be too long until we can get basecalling via this fast route," he predicted.
Loose said that he plans to continue to improve the Read Until method as well as develop applications for the method. "We're seeing how far we can push the envelope on Read Until," he said.
The main improvement to the method will be to "move from squiggle space to base space," he said. Instead of converting reference genomes into a current trace and basing decisions on whether to continue to sequence on matches to the current, he wants to be able to do basecalling in real time. Working in squiggle space, he said, restricts the potential analysis to smaller genomic regions of no more than about 5 Mb.
In addition, he said, the group is working on specific applications for the method, such as identifying pathogen sequences from a mixed background of primarily human DNA. Such a method could be useful for groups who are working to develop clinical diagnostics based on metagenomic sequencing, like Charles Chiu's group at the University of California, San Francisco. Another potential application would be genome finishing, Loose said. For instance, researchers often turn to long-read single-molecule sequencing technologies to sequence regions tricky to capture with short read technologies, such as repetitive regions. The Read Until technique could potentially be used to sequence regions like that in order to finish genomes.
Loose's lab has also begun using some of Oxford Nanopore's new products, such as the new R9 pore and flow cell, as well as the Rapid 1D library kit. He said that although he would love to test a Promethion, his group does not have plans in the near term to purchase one because the applications he is focused on do not require the throughput that a Promethion would deliver.
Loose said that the new R9 chemistry has resulted in approximately 5 percent to 6 percent better accuracy. He said it is unclear, however, whether the accuracy improvements are due to the pore itself or the fact that the company has also improved its basecaller, which uses neural networks rather than a hidden Markov model.
He also noted that accuracy is highly dependent on the quality of the input DNA. "The fresher the DNA, the better the quality of sequencing," he said, making it difficult to get a good measure of the baseline accuracy rate. "In our hands, it's extremely variable, depending on the quality of the DNA, but there is a definite improvement" with the R9 pore and new basecaller.
Loose said that at the firm's London Calling conference in May, some users reported issues with the flow cells, saying that they did not last as long as expected, limiting throughput. Loose said that his lab has experienced some issues with flow cells "dying off faster than they should," but that the flow cells' life continues to improve. Some users have reported generating between 2 Gb to 3 Gb of data, but he said his lab has yet to get 2 Gb from a single flow cell. He said he wasn't sure why the flow cells were not lasting as long as they should.
His lab has also been testing the recently launched Rapid 1D library kit, a two-step library prep protocol that takes 10 minutes to prepare genomic DNA to load onto the MinIon. "We've only done a few runs using that, but it is incredibly simple," he said.
Although the kit generates 1D reads only, Loose said that the advantage of it is its speed, simplicity, and the fact that it only requires 200 nanograms of input DNA.