NEW YORK (GenomeWeb) – As NanoString Technologies is preparing to launch its Hyb & Seq sequencing platform to early-access customers next year, it is starting to highlight potential clinical applications for the system.
The company believes Hyb & Seq may offer clear advantages over existing platforms in certain areas, including rapid detection of infectious pathogens and testing of their antibiotic susceptibility and resistance.
During a conference call last week discussing the company's third quarter financial results, NanoString CEO Brad Gray likened one type of use of the platform to a "supercharged nCounter system," which could "open up new clinical applications for NGS where existing systems fall short of [users] needs."
He added that the ability to support rapid infectious disease detection and drug susceptibility testing could be one of the "killer applications" for Hyb &Seq, a sequencing-by-hybridization platform in which DNA or RNA is read six bases at a time through the use of thousands of unique hexamer barcodes.
Earlier this month at the annual meeting of the Association for Molecular Pathology in San Antonio, the company and early-access customers presented early data from the platform.
Joe Beechem, NanoStrings's senior vice president of research and development, said at the company's corporate workshop at AMP that the firm now has a "full-blown general-purpose sequencer up and running" that uses a set of 4,096 probes.
Previous data the company has presented was generated using a smaller set of probes than the 4,096 necessary to provide complete coverage of any unknown genome thrown at the system. Now, "it doesn’t matter what kind of DNA or RNA [you are working with,] it's enough to sequence anything," he said.
NanoString has sequenced billions of bases by now, showing that when it covers each single base with at least three barcodes, it can reduce the error rate to a very low level.
The platform also generates the same type of files as any other commercial sequencing now, which can be input to a standard genome browser or NGS viewer, he pointed out.
As the company looks toward establishing the platform for clinical applications, Beechem also stressed the ability of Hyb & Seq to be used for counting applications, as a kind of intermediary between the firm's existing nCounter technology and true sequencing, a use that requires "ultra-fast molecular barcoding."
"With sequencing-by-synthesis, if you want to count an RNA, you have to sequence and then order it and count," he said. "But we can use barcodes designed for counting."
This can happen as much as forty times faster than sequencing with the full 4,096 barcodes, in timeframes as low as two and a half hours, he added.
Thus, the same system and chemistry can serve various applications, including ultra-rapid targeted counting for infectious disease testing, sequencing of small panels for oncology in as little as nine hours, sequencing of larger panels within about a day, and whole-genome sequencing.
Roby Bhattacharyya, a researcher in the Division of Infectious Diseases at Massachusetts General Hospital, has explored the Hyb & Seq platform for rapid infectious disease testing. He and his colleagues have mostly been using NanoString's older nCounter platform so far but are now replicating and expanding those results on the Hyb & Seq.
"Standard diagnostics do a good job of telling us we are dealing with a resistant organism," Bhattacharyya said during the NanoString corporate workshop, "but they do it too slowly for us to use that information to choose [which antibiotics to use] … so we end up guessing, at least for the first few days."
Assays using MALDI mass spectrometry to identify organisms have already been developed, and whole-genome sequencing has become a technical gold standard, if not a clinically-implemented one, he added.
In one arm of their research, Bhattacharyya and colleagues have been using the nCounter initially, and now Hyb & Seq, for their so-called phylogeny-informed ribosomal RNA-based strain ID (PhIRST-ID).
In initial experiments, the team was able to reach 90 percent agreement with established methods at the species level, and 100 percent at the family level, at fewer than 100 bacteria in a sample — more than enough for sputum or urine and much more than necessary for blood cultures. "But it doesn't quite get us to uncultured blood," he said.
Using the new Hyb & Seq platform, Bhattacharyya said he and and his colleagues can get 100 times more counts from the same sample.
"This gets us where we don't think it's unreasonable to get to direct identification from uncultured blood, which as we know from other applications, can be absolutely transformative," he said.
"Our standard so far has been three and a half hours for a single sample and seven and a half for batches of 12," he added. "But with Hyb & Seq, everything goes faster, so our initial goal [with that] is three and a half hours for a panel of eight."
In addition, "getting down to the base level with [the new system], there is no reason you can't get higher accuracy at the species resolution, as well," Bhattacharyya said. Potentially, a coarser initial approach could be used upfront to identify a smaller region in which to do resequencing using the Hyb & Seq in order to identify species unambiguously with less coverage, an approach the team plans to pilot.
Identifying organisms responsible for an infection is only one step, though, Bhattacharyya said. The other is determining what drug to treat the patient with.
Available tools, based on either sequncing or PCR, can already test for genetic loci associated with antibiotic resistance, but this is not enough for a clinician to choose an antibiotic, he argued.
"What we really need is to know susceptibility, not just resistance," he explained.
In this vein, he and his colleagues have been developing rapid phenotyping using RNA, based on the hypothesis that the counting of transcripts could provide a measure of a pathogen's response to drug exposure.
"We needed a fast and sensitive way to quantify multiple transcripts in a single assay, but simple enough to think about doing it on patient samples repeatedly, so that's where we met up with NanoString," he said.
The group calls their method PhAST-R (phenotypic susceptibility testing through RNA-detection).
"Transcriptional changes are among the earliest change in response to cell stress, so our postulate was that cells that are susceptible or resistant will have different transcriptional responses to antibiotic exposure," Bhattacharyya explained. "A cell that is dying will look much different than a cell that is not dying."
To begin, the team generated expression signatures using RNA-seq data, focusing on three antibiotics. As they hoped, they saw that susceptible strains had a dramatic rearrangement of their transcriptomes within an hour of drug exposure, which resistant strains did not.
Using machine learning, the investigators were able to boil the expression signatures down to just 10 predictive genes, which showed a strong ability to distinguish susceptible and resistant organisms, with some exceptions.
In addition to providing a measure of susceptibility that is independent of established genetic resistance mechanisms, the nice thing about RNA is that it can give you a parallel readout on genotype, Bhattacharyya said, which allows cross-referencing of phenotypic susceptibility and genotypic resistance.
In one discordant case, where the transcription change predicted resistance but the organism was known to be susceptible, the team was able to identify a genomic change that confers a strong inoculant effect, suggesting that the resistance call was actually correct.
Testing 250 isolates, the group had 96 percent agreement with standard testing overall, and most miscalls were cases where the method called a bug resistant while standard assays called it responsive.
"I'd argue that is the kind of error you'd rather make," Bhattacharyya said.
With older NanoString technology, he said, the method works from a blood culture bottle in about four hours. "But again," he said, "we can get faster with Hyb & Seq."
In the group's early studies with the new platform, it is seeing similar accuracy, he added.
Bhattacharyya and colleagues are not the only group that has tried the Hyb & Seq platform in the infectious disease space. Another group, led by associate professor Chris Mason at Weill Cornell, shared a poster at the AMP meeting on methods they have described previously for cross-kingdom pathogen identification, using both DNA and RNA.
The team designed a sequencing panel covering 10 species, including yeast, gram-negative and gram-positive bacteria, and an RNA virus. It interrogates 18 to 20 genomic sites per species, with additional ribosomal RNA probes for five of the bacterial species to allow simultaneous DNA and RNA detection.
The investigators tested clinical samples and compared results obtained using the Hyb & Seq technology with pathology findings, as well as qPCR tests run in parallel, showing full agreement in all cases.