NEW YORK – Nanopore sequencing is no longer just for long reads. Now, under the right circumstances, it could make sense to use it for sequencing molecules as short as 400 bp.
Researchers from Memorial Sloan Kettering Cancer Center and Johns Hopkins University have come up with a strategy for low-coverage copy number analysis in cancer cells using Oxford Nanopore Technologies' platform.
"Normally with nanopore sequencing there's concern about base accuracy, but that's really not relevant here," said Michael Schatz, a computational biologist at Johns Hopkins. "The way you find these copy number variants with sequencing is primarily a counting exercise."
By dividing up a large segment of the genome into bins, one can align shorter reads and simply count how many are aligned to each bin. "You get a very robust estimate where you can detect amplifications or deletions, heterozygosity, or even larger scale copy number alterations," he said. But one needs lots of independent reads, millions of them. "At minimum, you'll need thousands of reads in each bin," he said.
Schatz's team hypothesized that loading the ONT platform with short fragments of DNA would simply yield many more reads, enough to do their low-coverage studies at 1X or even 0.1X coverage. A paper published last month in Nucleic Acids Research shows that not only were they right, their method compares favorably with copy number analysis by Illumina short-read sequencing and even fluorescence in situ hybridization, the prevailing method for copy number analysis.
They were able to detect copy number events ranging in size from a few mb to a whole chromosome arm and chromosome-level events, including focal amplifications at the MYC and ERBB2 genes.
"The short version is we can detect all the variants observable with FISH," Schatz said. "Plus, it's cheaper, faster, and more reliable. There are a few samples you just couldn't get a readout with FISH, but we could get a readout with a sequencing assay."
Schatz said he does not have plans to commercialize the assay but is keen to develop it further. "We're trying to do this systematically."
The approach has been in development for several years, Schatz said, though it was "punctuated" due to the COVID-19 pandemic. It represents the merging to two interests: cancer CNV profiling and nanopore sequencing technology development. Before working with nanopore sequencing, his lab developed a computational suite for single-cell CNV profiling of breast and other cancers, called Ginkgo.
FISH is the clinical standard for CNV analysis, which is important to cytogenetics, but it requires microscopes and technicians, and there is a limit to how big or small the CNVs can be. "The field, as a whole, is moving towards sequencing" with Illumina's platform, he said. But it comes with added library preparation expenses.
To show proof of concept, the researchers first had to show that the ONT platform would work with shorter molecules. They loaded sonicated libraries of molecules with an average length of approximately 500 bp onto a MinIon device, yielding up to 6 million sequenced molecules in a single run. The authors noted that these reads were of similar quality to long reads when comparing sequencing metrics such as pass filter, approximately 85 percent for both, and mean Q-score values — 10.54 for short and 10.17 for long.
To demonstrate clinical potential, the team analyzed five acute myeloid leukemia samples, using both nanopore and Illumina platforms. All samples yielded high read counts of 4 million to 6 million reads and the data yielded "largely similar genome-wide profiles and CNAs … when compared to Illumina data." One sample inferred copy number aberrations with large deletions on chromosomes five and seven, complex rearrangements on chromosome 11 resulting in gene gains, similar breakpoint positions, and even focal alterations — corroborated with FISH.
Other experiments showed that the method could work with multiplex samples and, in even more constrained circumstances, with ONT's Flongle, a sequencing device that is cheap and portable but offers limited sequencing yield compared to MinIon.
Now, the team is ramping up to do larger studies. Schatz said he plans to merge this method with his work on bioinformatics-based adaptive sequencing, a unique feature of ONT's platform. In regions of interest, one could get deeper coverage and potentially longer reads, to provide info on structural variants.
"The idea is you'd simultaneously do a genome and CNV," he said. "If there are parts of the genome where you want deeper coverage, you can do on-the-fly enrichment of that."