Accelerated identification of disease-causing variants with ultra-rapid nanopore whole human genome sequencing reported in full in Nature Biotechnology

Oxford Nanopore worked with a team led by Stanford University to show the potential value of rapid whole genome sequencing for use in critical care settings, where actionable insights can be gained in hours instead of days

In a landmark study published in Nature Biotechnology today, a team of researchers from Stanford University School of Medicine describe how they developed a workflow for ultra-rapid nanopore sequencing that resulted in actionable characterisation of genetic disease in under 8 hours. The work used Oxford Nanopore’s high-throughput sequencing device, PromethION 48, which is uniquely designed to enable accelerated sequencing by using multiple flow cells to run one sample.

12 seriously ill patients, ranging from three months old to 57 years, were chosen by a team of researchers from Stanford University School of Medicine to take part in this study, all of whom presented with critical illness and hard to diagnose symptoms. The team set about cutting the time for a whole human genome sequence in collaboration with Oxford Nanopore Technologies, NVIDIA and Google. For all study participants, the full ultra-rapid pipeline was completed within a single day and pathogenic genetic variations were identified for five of the 12 patients. This represents a diagnostic yield of 42%. The study demonstrated that rapid identification of pathogenic variants has the potential to guide physicians in clinical management, improve prognosis, and reduce treatment costs.

Standard genetic tests can take days or even weeks to be returned and may not give all the information that is required for physicians to make clinical inferences. In one of the five patients (previously described in the NEJM) the standard panel of genetic tests were returned two weeks later and were inconclusive, compared to rapid nanopore sequencing that provided actionable insights helping to enable physicians to determine the cause of this infant’s epilepsy in 8.5 hours. Using a whole genome approach enables researchers in this situation to proceed without a hypothesis and nanopore sequencing reveals rich data that includes many types of genetic variation, which is a significant advantage compared to current methods.

Two of the 12 patients are discussed in this paper, the first being a 57-year-old male with many medical issues including hyperthyroidism, hypertension, severe COVID and requiring a lung transplant. The team of doctors were conducting myriad tests, however rapid nanopore sequencing uncovered a genetic variant that physicians determined was “likely pathogenic”, reducing the need for further tests and a heart biopsy.

Gordon Sanghera, CEO, Oxford Nanopore Technologies commented:

“We are delighted to see this pivotal work published in full, and to show the potential impact that ultra-rapid nanopore sequencing can have for critically ill patients. The research showed that compared with some standard diagnostic methods that can takes weeks, this provided accurate insights for clinicians to inform their decision making and enable them to give answers to their patients and their families all in the same day.

60X coverage in 2.5 hours equates to a whole human genome being sequenced every 2.5 minutes and we believe we can streamline this to cut the time in half. This work illustrates a future where rich, fast genomic insights could fundamentally change critical care.”

Find out what one of the lead researchers, John Gorzynski, has to say about this on Twitter:

A framework to provide accurate variant calls and prioritisation

Today’s publication in Nature Biotechnology outlines in detail how the team used PromethION 48, Oxford Nanopore’s highest-throughput sequencing device, to run a single sample in parallel across all 48 flow cells on the device and deliver a whole human genome sequence in just two hours of sequencing.

Small variant calling performance with and without barcoding was shown to be alike and similarly robust outcomes were observed in the barcoded and non-barcoded structural variant (SV) calls. Based on this data, the team elected to continue without barcoding. This resulted in a 37-minute reduction in total library preparation time and improved downstream sequencing efficiency.

Speed did not impact accuracy of variant calling, with the team demonstrating an F1 score of 0.9974 for SNPs and 0.7322 for INDELS benchmarked on HG002. Variant calling resulted in a median of 4,490,490 small variants, and 22 prioritised structural variants per sample.

Real-time basecalling and alignment was addressed by developing a cloud compute infrastructure (Google Cloud Platform) and parallelised basecalling and alignment across multiple GPU instances (NVIDIA GPUs). Using these methods basecalling and alignment of a high depth (200 Gb), long-read whole human genome was completed in near real-time.

Finally, to accelerate variant calling, it was scaled to several cloud instances and genomic sections were run in parallel with a varying number of threads resulting in a runtime of 29 minutes.

Separate research has shown that 34% of all disease-causing variation is made up of variants that are larger than a single base-pair substitution1, therefore as nanopore long-reads reveal greater insight and more information, these capabilities provide significant benefit. A further advantage of nanopore sequencing in this instance was the ability to derive methylation information directly from the raw signal. While methylation changes as a cause of genetic disease are uncommonly recognised, the authors hypothesise that literature reports are likely to be a significant underestimate.

The estimated cost of this ultra-rapid approach at scale including DNA extraction, library preparation, sequencing, and computation is just over $5000 including $1600 in flow cell costs, $1584 in Oxford Nanopore reagents and $568 in compute costs. Oxford Nanopore expects these costs to reduce substantially with improvements to sample prep, compute and flow cells.

The full paper can be accessed for free here: https://www.nature.com/articles/s41587-022-01221-5

-end-

Reference:

  1. Genetic Variation, Comparative Genomics, and the Diagnosis of Disease. Evan E. Eichler, July 2019. N Engl J Med. 2019 Jul 4; 381(1): 64–74.

Forward-looking statements

This announcement contains certain forward-looking statements. Phrases such as "potential", “expect”, "intend", “believe we can”, “working to”, "anticipate", "when validated", and similar expressions of a future or forward-looking nature should also be considered forward-looking statements.  Forward-looking statements address our expected future business, and by definition address matters that are, to different degrees, uncertain and may involve factors beyond our control.