NEW YORK – Following a high-profile online event last month that revealed its Sequencing-by-Expansion (SBX) nanopore sequencer, Roche cast a spotlight on the platform at this year's Advances in Genome Biology and Technology (AGBT) annual meeting with testimonies from the early-access collaborators.
Pricing for the sequencer, which Roche plans to launch as a research-use-only product in 2026, remains an open question, though.
So far, Roche has deployed early versions of the SBX platform at two early-access sites: the Hartwig Medical Foundation in the Netherlands and the Broad Institute. Researchers from both institutions presented their experience at a Roche-sponsored workshop during AGBT last week.
Edwin Cuppen, Hartwig’s scientific director, said his team received a prototype SBX instrument from Roche last September. Leading the foundation's efforts to provide whole-genome sequencing-based cancer diagnostics in the Netherlands, Cuppen said routine implementation of WGS in the clinic is often limited by two bottlenecks: cost and speed. "This is why we are always excited to explore new technologies to see if we can address these challenges," he said.
Following preliminary test runs and optimization steps, Cuppen's group commenced the first production run last December with six patient samples and controls. In January, the team performed the second production run with 30 tumor-normal pairs and two COLO829 cell line controls.
For the January experiment, the Hartwig team split the samples into eight sequencing runs using a single array chip. Each sequencing run lasted four hours, Cuppen said, and with the help of automation, turnaround time for the entire workflow, from sample to data analysis, was 24 hours.
Overall, Cuppen said the sequencing runs yielded 150 Gb of data for the normal samples and 410 Gb of data for the tumor samples. The coverage was "fairly good," and there was not much variation between samples, Cuppen noted. Of the total reads, 70 percent were full duplex reads, he said, with an average N50 (a metric for evaluating read length) of 190 bp. For the 30 percent partial duplex reads, the N50 was 110 bp.
Cuppen's team also compared the SBX data with matching Illumina NovaSeq 6000 data generated for the same samples from routine diagnostics. Overall, the comparison showed "highly similar" genome coverage, Cuppen said, with coverage for the tumor samples being almost identical between the platforms and coverage for the normal tissues being "slightly lower" on SBX.
Looking at GC bias, Cuppen said the team observed SBX coverage dropping off earlier than Illumina at AT-rich regions. Meanwhile, for regions with high GC content, performance of SBX was "much better" than Illumina, he noted.
The Hartwig researchers also performed somatic variant calling and cancer genomics analysis on the SBX data using their in-house-developed pipeline Hartwig WiGiTS. Comparing the data with corresponding variant analysis data from Illlumina, Cuppen said the two platforms showed "highly concordant" small variant calls, "near identical" copy number profiles, and "good concordance for structural variants."
Judging by the preliminary results, Cuppen said the diagnostic performance of the SBX platform "would be highly similar to Illumina-based routine diagnostics."
Similar to the Hartwig team, the Broad Institute is also testing out SBX for potential clinical applications. "One of the biggest places that we immediately thought that we could unlock potential is using this technology for rapid whole-genome [sequencing] in the NICU," said Sean Hofherr, chief of clinical strategy and product development at Broad Clinical Labs (BCL).
According to Hofherr, BCL received the SBX prototype platform in January. Benchmarking the platform with four Genome in a Bottle reference samples, Hofherr said the overall performance of the instrument was "really solid," and the quality metrics for SNVs and indels were "extremely impressive."
The BCL team further tested the platform with human cell lines from the Coriell Institute for Medical Research that have known single nucleotide variants, copy number variants, and repeat expansions. In each of these cases, the causative variant was successfully identified, Hofherr said.
Going a step further, Hofherr's team tested the platform with existing clinical samples that had been diagnosed with a known condition by BCL and ran the variant analysis using the automated AI platform developed by Fabric Genomics. "We were able to detect the positive variant that we reported in our laboratory for each of these patients," Hofherr said, adding that the SBX data worked "perfectly without any issue" with the Fabric pipeline.
In terms of the turnaround time, Hofherr said using the rapid SBX workflow, his team was able to go from sample to results for a sample in six hours and 25 minutes.
Broad researchers led by Aziz Al'Khafaji, leader of Broad's methods development lab, also evaluated SBX for RNA research applications. His team is eyeing SBX for two applications: multiomics drug response profiling using multiplexed cell line screening and single-cell RNA isoform sequencing.
For the former, Al'Khafaji said his team ran 10x Genomics Flex single-cell libraries on SBX, and data so far has been "phenomenal" with "no obvious batch effect detected." While the RNA isoform analysis is still ongoing and at an early stage, Al'Khafaji said, early results have shown SBX can generate a range of read lengths, which "gave us a lot of confidence that these read lengths could span multiple splice junctions and even resolve full-length isoforms."
While Roche's roadshow appeared to have impressed researchers, the company still has not addressed the elephant in the room: pricing.
During the workshop, Gustav Karlberg, Roche's VP life cycle leader of sequencing systems, acknowledged that pricing is one of the most frequent questions the company gets for SBX.
"We know what the market expectations are for high-throughput [sequencing]," he said. "We know that this technology will and can meet all those expectations when it comes to consumables and run costs."
"The accuracy so far is good by whatever they have shown," said Xiaoling Xuei, a sequencing consultant and retired technical director of Indiana University School of Medicine's Center for Medical Genomics, who attended the Roche workshop. Judging by the presented data, "what is good about this technology is, it is fast, and it has decent read length."
Still, a big question for her is the cost. "It really comes down to the pricing," including reagent and chip costs, she said.
"The equivalence to Illumina as a diagnostic platform for WGS was impressive, as well as the high reported Q scores and data throughout," said Christopher Mason, a genomics expert and professor at Weill Cornell Medicine. "The big questions now include how it will perform in 200 labs versus two, the cost framework by box and by base, and the GC variation," he added.
Other researchers still have reservations about the SBX technology's reliability and operability, given the many new concepts Roche has introduced.
"I have doubts on how robust it will be," said Catharine Aquino, head of the genomics core facility at the Functional Genomics Center at ETH Zurich, who also attended Roche's workshop. "The Broad can make everything work … I will only believe that it actually is robust if I touch it."