Skip to main content
Premium Trial:

Request an Annual Quote

Repeat Expansion Disease Diagnosis From Targeted Long Reads Shows Potential for Clinical Use


This article has been updated to correct the affiliation of Danny Miller. 

BALTIMORE – Using programmable targeted long-read nanopore sequencing, researchers from the Garvan Institute of Medical Research in Sydney and their collaborators have devised a test that can detect all currently known short tandem repeat (STR) expansions associated with rare neurological diseases, suggesting a potential path to end the diagnostic odyssey of patients afflicted by one of those disorders.

The assay, described in a proof-of-concept study published last week in Science Advances, tapped Oxford Nanopore Technologies' long-read sequencing workflow and its recently released selective sequencing software feature. It is capable of discerning all known neuropathogenic STRs, implicated in more than 50 neurological and neuromuscular diseases, across dozens of genes while simultaneously providing DNA methylation profiles, thus enabling the diagnosis of challenging genetic disorders.

"We chose to focus on these short tandem repeat disorders for a couple of reasons," said Ira Deveson, head of genomic technologies at the Garvan Institute and the study's senior author. Among them, he said, is that STR expansions, which make up around 7 percent of the human genome, can cause a plethora of heritable disorders, including Huntington's disease, fragile X syndrome, and hereditary cerebellar ataxias.

Another reason, he added, is that despite their wide variety and collectively high prevalence, STR expansion disorders are still hard to diagnose clinically. According to Deveson, there are currently a few "old school molecular biology techniques," such as Southern blot and repeat-primed PCR, to help identify STRs for specific disease phenotypes. However, these methods can typically only target one gene at a time. Because many STR expansion disorders have overlapping symptoms, "you can end up in a situation where the patient has to go on this long diagnostic odyssey of having multiple different tests," he said.

Even though next-generation sequencing presented some promises to detect STR expansions, the large size and highly repetitive nature of STRs have made them difficult to be solved with short-read sequencing methods, such as Illumina sequencing, Deveson noted, although researchers in the UK recently employed whole-genome Illumina sequencing and new software to detect a variety of repeat expansion disorders.

The repeating sequences are typically thousands of bases long, and short-read sequencing is trying to stitch them together from fragments of hundreds of bases, he explained. "It's kind of like trying to do a jigsaw puzzle where all of the pieces look the same."

To solve the problem, Deveson's team used Oxford Nanopore sequencing for its test. One of the "really exciting" aspects of ONT sequencing is that it has "no upper limit on the read length," he said. "[T]hat allows you to sequence straight through these long, repetitive sequences and directly read or solve them without any dramatic challenges."

For their study, the researchers analyzed genomic DNA samples from 37 participants, including 25 patients with various neurogenetic diseases and 12 controls without disorders. The goal of analyzing these patients, who have "fairly clear neurological disorders" and have also been "subjected to the standard molecular testing paradigm," Deveson said, was to show the test's ability to diagnose STR expansion disorders as successfully as traditional testing.

Importantly, the assay deployed Oxford Nanopore's ReadUntil software, a recently released adaptive sampling feature that allows the platform to target specific DNA fragments. ReadUntil can "very quickly recognize a DNA fragment as it enters the pore," Deveson explained, and if it seems of no interest, it can "stop the sequencing on that pore and spit that fragment out," making the sequencing "much more efficient and effective." Moreover, compared with other wet-lab-based enrichment methods, such as using Cas9, ReadUntil does not require any additional laboratory procedures. "The device itself is actually doing all of the targeted gene selection in a programmatic fashion," said Deveson, "which is really pretty amazing."

The authors directed ReadUntil to target a total of 37 genes known to harbor STRs associated with neurological and neuromuscular diseases. Using haplotype-resolved assembly and DNA methylation profiling of STR sites in the sequencing data, the assay successfully identified all individuals with a long list of STR expansion disorders, including Huntington's disease; fragile X syndrome; cerebellar ataxia, neuropathy, and vestibular areflexia syndrome (CANVAS); spinal-bulbar muscular atrophy; myotonic dystrophy type 1; neuronal intranuclear hyaline inclusion disease; Friedreich's ataxia; amyotrophic lateral sclerosis; spinocerebellar ataxia type 1; and oculopharyngeal muscular dystrophy.

The results showed that the test is "at least as good" and in fact "appears to be quite a bit better" than existing molecular methods, said Deveson, noting that it was even able to pick up STR alleles in some pathogenic genes — such as RFC1, which is associated with CANVAS — that were unsolvable by traditional molecular methods or short-read sequencing.

Also, with ReadUntil, the team was able to achieve similar coverage for the targeted genes on a MinIon sequencer as with whole-genome sequencing on a PromethIon, ONT's highest throughput platform. According to Deveson, the study reported about a fivefold sequencing enrichment with ReadUntil, "which effectively makes your assay five times cheaper per data unit." Furthermore, because ReadUntil is programmable, the team was able to add 28 pharmacogenomic genes as second targets to help predict a patient's drug response and treatment prognosis, demonstrating the assay's flexibility and compatibility with multi-pronged targeted testing.

"I really liked the paper," said Danny Miller, a resident physician in the combined pediatrics and genetics program at Seattle Children's Hospital. Because the assay can offer "a lot of information" in one take, he said, "it shows the utility in both the research setting and, hopefully, in the clinical setting soon."

Miller, a physician-scientist, and his team are also using Oxford Nanopore sequencing to solve hard-to-diagnose pathogenic variants within the human genome sometimes with the help of the ReadUntil feature. He thinks that this study represents "a really nice first step" to demonstrate long-read sequencing's utility in STR expansion disorder testing, though "a lot of open questions" still need to be sorted out.

For one, Miller said that he is "still not convinced" that the ReadUntil function will be more efficient than whole-genome sequencing for a clinical lab. Although he acknowledged the theoretical cost advantage of using ReadUntil, he said that the feature is still more likely to require "additional hands-on time" since it "doesn't work perfectly every time," affecting its competitive edge.

Another potential issue is the variability of the expansion between individual sequence reads, Miller said. He explained that in his own research, his team has seen variability of repeat copy numbers among different sequencing reads on the same targeted repeat expansion. If that happens, "[w]hat do you report, and how many reads do you need to report something that you trust?" he asked. "I don't think we know that yet." For the test to be adopted in clinical labs, he said, each lab would need to "come up with their own guide to figure out, 'here's what we need to see in order to call this a successful experiment that we will sign off on.'"

Additionally, given homopolymers and indels are "the biggest errors source for nanopore [sequencing]," Miller said, people need to be "really careful when we look at a repeat expansions, or any repeat unit in the genome, using nanopore [sequencing]" and "have some type of systematic evaluation" to validate the accuracy of the test.

Tomi Pastinen, director of the genomic medicine center at Children's Mercy Hospital in Kansas City, also said that the study has a lot of strengths. "Previous studies have looked at individual repeats and not across many different repeats," he said. "This study uses multiple positive controls from known disease cases and showed good sensitivity and specificity for detecting the repeats." Pastinen's group is also investigating long reads to tackle tricky genetic disease cases. However, instead of using nanopore sequencing, his team mostly works with Pacific Biosciences HiFi sequencing.

Despite the advantages of long-read sequencing for analyzing STR expansions, short-read sequencing paired with software has also demonstrated the ability to outperform conventional testing methods for detecting neurological repeat expansion disorders. For instance, a recent paper published in The Lancet Neurology by Illumina and UK researchers showed that with the help of ExpansionHunter, a sequence-graph-based tool to analyze variation in short tandem repeat regions developed by Illumina, whole-genome Illumina sequencing showed 97.3 percent sensitivity and 99.6 specificity for detecting repeat expansions across the 13 most common repeat expansion loci, compared to conventional PCR testing.

ExpansionHunter "is getting better and better" for detecting repeat expansions with short-read sequencing, Pastinen said. "But to my knowledge, that is not an extensive diagnostic used to date."

He pointed out that one of the weaknesses of the new nanopore test is that "it is still in the same mindset" as individual repeat disorder tests since it is "a targeted test" and therefore becoming "a separate test that [clinicians] would order in parallel to genome sequencing."

Another limitation of the test, he said, is its "restriction to known repeat expansion diseases." Because the test requires researchers to program ReadUntil to look for specific pathogenic gene targets, the detection is limited to existing expansion disorders that have been previously defined and studied.

Similar to Miller, Pastinen also said that although the ReadUntil function is "a really nice feature," he is not sure about its use in a standard clinical lab setting since a test's simplicity and turnaround time are the "more important things" for patients and providers.

Deveson said the current turnaround time for the test is about 75 hours for library prep and sequencing plus around 24 hours for bioinformatics analysis. With that, he thinks a turnaround time of a week for this test is definitely "comfortable."

He also acknowledged that the ReadUntil approach would not be suitable for a gene discovery project, where high-throughput whole-genome nanopore sequencing "would be the way to go."

Moving forward, Deveson said his team is interested in pursuing clinical accreditation for the test. To achieve that, the next step would be "to analyze a larger cohort of deeply characterized reference samples." Although he is optimistic that the test will go through a similar clinical validation process as short-read sequencing, he said one hurdle he can foresee with nanopore sequencing "is the fact that the technology is really developing quite quickly" as it is a fairly recent technology. "New updates to the nanopore devices, the sequencing chemistry, and the software happen all the time," he said. "But from a clinical validation and accreditation perspective, you need to lock all those things down and stick with the current version — you can't keep changing every few months."

Future research will likely recruit participants with STR expansion disorders that weren't included in the current study, he added, "simply because we just want to make sure that the widest possible range is covered." Eventually, he hopes to deploy the method to develop similar tests for not just neurological repeat disorders but "any other genetic disorders where there's a challenging mutation involved."

As for the potential commercialization of the test, Deveson said he and collaborators "don't really see this [research] as a business venture" and don't think there is anything of interest from an IP perspective. Instead, he said, "we're primarily interested in making a new test that works and will improve diagnosis of people with these complex disorders."