Skip to main content
Premium Trial:

Request an Annual Quote

Oxford Nanopore Technologies Unveils New Direct RNA Sequencing Chemistry


NEW YORK – Oxford Nanopore Technologies said on Thursday that it is rolling out a new direct RNA sequencing kit that the company promises will "significantly" increase the raw read accuracy as well as data output.

While Oxford Nanopore has been known for its relatively rapid product turnover for DNA applications, the new RNA kit, anticipated by many researchers, appears to be the first major overhaul to its RNA sequencing chemistry in several years.

According to Libby Snell, a lead scientist at Oxford Nanopore who oversees RNA-based sequencing, the new RNA sequencing kit, Sqk-RNA004, is somewhat of a skip-generation update to the current RNA002 kit, which was initially released around 2018.

"At one point, we were working on chemistry called RNA3," Snell told attendees during the company's annual user meeting in London this week, which was webcast and where the new RNA chemistry was announced. "We were not quite comfortable with actually releasing it. We took a step back and reassessed and started working on a different chemistry, which is RNA4."

Unlike the current RNA kit, which utilizes the same pore as DNA sequencing, a key feature of the new chemistry is that it employs a dedicated pore for RNA sequencing named RP4, Snell said. This means the new kit will require different flow cells from those for DNA sequencing.

In addition, Snell said the new RNA chemistry will be equipped with a faster motor, translocating the RNA molecule at about 120 bases per second, compared to the current speed of 70 bases a second. With this data output, Snell claimed that an RNA sequencing run can deliver upward of about 30 million reads per PromethIon flow cell after a full run.

The updated chemistry also promises to deliver improved raw read accuracy by incorporating a larger neural network and expanded datasets for training, Snell said. RNA004 is expected to reach about 96 percent single-read accuracy for the human transcriptome, "a significant improvement" compared to the current version, she said.

When it comes to sample prep, Snell said the RNA004 kit will have a "very similar" protocol to RNA002. In both cases, the workflow starts with 3' ligation, followed by an intermediate reverse transcription step to generate a DNA-RNA hybrid molecule. However, Snell said the reverse transcription product is only to provide a scaffold to help the native RNA molecules translocate through the pore, and the cDNA strand will not go through the pore or be read by the sequencer.

As for input requirement, Snell said the new chemistry calls for between 100 ng and 200 ng of enriched RNA or 1 μg of total RNA as starting material. "You can have lower inputs, you just have to understand that you are going to get less output," she added.

While public data for the RNA004 kit remains scant, Snell teased the audience with some internal benchmarking results. When tested with quantitative RNA transcript panels, the RNA004 showed a "really high correlation" between expected versus observed concentrations of target RNA molecules, she said. Additionally, she noted that there was "a very nice correlation" between samples sequenced with the RNA002 and RNA004 kits, and the new chemistry showed consistent data quality in technical replicates.

Martin Smith, a computational biologist at Sainte-Justine University Hospital Research Centre in Canada and one of the early users of the RNA004 kit, shared his initial impression of the new product in a meeting presentation. His team, which focuses on studying long noncoding RNAs, received the new RNA flow cells only about three weeks ago, he said.

The team first tested the new chemistry using commercial poly-A enriched RNA, generating close to 15 million reads on a PromethIon flow cell. When applied to total RNA, the kit generated roughly 14 million reads. Smith said he was impressed by these results, adding that "this is the update we were waiting for."

He noted that the poly-A enriched RNA tended to drop the read length a little bit, possibly due to the extra manipulation step required for the poly-A selection protocol. In terms of data quality, the group achieved 95 percent to 96 percent single-base accuracy. While the known homopolymer issues persist in the new chemistry, Smith said, he believes they can be addressed with raw signal analysis.

After the initial benchmarking, Smith and his team used 20 RNA004 PromethIon flow cells for functional cancer genomics analysis. Specifically, his team generated native transcriptomes from Nalm6 cells with CRISPR-Cas13 knockdowns of four cancer-associated long noncoding RNAs identified through a CRISPR-Cas9 screen. Overall, Smith showed that the new kit was able to help identify new RNA isoforms and new long noncoding RNAs in the samples.

"I think the main improvement of the new kit is the improved throughput, which is a very nice feature for the whole community," Eva Maria Novoa, a group leader at the Center for Genomic Regulation in Barcelona who is an expert in nanopore RNA sequencing, said in an email.

Novoa said her group, which recently published a new method for nanopore sequencing of tRNAs, has not yet tried the new RNA kit. However, based on what the company presented as well as personal interactions with some of the early users, she thinks Oxford Nanopore's promise of a three- to fourfold throughput increase might be "over-optimistic."

Nonetheless, she believes the improvement in sequencing yield will allow researchers to explore a broader range of transcripts, and the enhanced coverage per transcript can also boost the statistical power to help researchers identify dysregulated transcripts through differential expression analysis.

While Novoa thinks the improved accuracy for RNA004 is "very nice," she also pointed out that this accuracy could already be achieved with the current chemistry with improved basecalling models. Her group, for instance, has already achieved a 96 percent to 97 percent raw read accuracy using the RNA002 chemistry with basecallers trained in-house.

Moreover, it remains to be seen whether the new chemistry will be better or worse at detecting RNA modifications, which can be important for RNA studies, Novoa pointed out. As the new chemistry involves a new pore, researchers are likely to have to retrain their existing basecalling models for RNA modifications.

MinKnow, Oxford Nanopore's proprietary sequencing data analysis, is still not able to call RNA modifications directly, but a company spokesperson said this feature will be developed "in due course."

During a separate technology update presentation, Clive Brown, Oxford Nanopore's chief technology, innovation, and product officer, pointed out the challenges of calling RNA modifications, of which there are at least 170 different types.

"RNA is often quite heavily modified," Brown said. "So, we've got the same problem we had on DNA, but worse." Company researchers have identified a handful of RNA modifications as a starting point, he added, and are synthesizing RNA libraries with modifications in known positions to help train the basecalling software.

In addition to RNA sequencing, Brown also offered updates on the company’s R&D pipeline, including a new MinIon model with iPad connectivity and new application specific integrated circuit (ASIC) chips, which will support a new family of lower-cost, lower-power devices.

According to Oxford Nanopore Senior VP of Product & Program Management Rosemary Sinclair Dokos, the RNA004 kits are currently available in develop access, followed by early access this summer as the company scales up the production of the kit.