NEW YORK – Genome technology company JumpCode Genomics emerged from stealth mode this week with a CRISPR-based technology that aims to turn down the background noise in next-generation sequencing libraries.
The technology, called CRISPRclean, works by depleting unwanted nucleic acid sequences from NGS libraries to leave behind only targets of interest. According to JumpCode, it has broad applicability in several fields, including oncology, infectious diseases, and undiagnosed genetic diseases and disorders, and it's agnostic to library prep methods and sequencing platforms.
The company, which was founded in 2016, also added new executives to its roster, including Stan Nelson, vice chairman of human genetics at the University of California, Los Angeles as consulting chief scientist; Gary Schroth, VP and distinguished scientist at Illumina, as a member of the company's scientific advisory board; and CEO Yaron Hakak, who recently served as director of business development at Illumina.
Advances in NGS have accelerated in the last decade, and as the technology gets faster and gets combined with other platforms like machine learning, cloud computing, microfluidics, and automation, the amount of data being generated through each sequencing run increases significantly, Hakak noted. The problem, however, is that most molecules in any given biological sample being sequenced are uninformative for any given application.
"It's the needle-in-the-haystack problem," he said. "You get a lot of noise blocking the signal you actually want to see. Aside from dramatically increasing your sequencing cost, it's limiting your ability to actually do what you want to do. And that's exactly where CRISPRclean comes in. What in essence it does is remove that haystack for you so you can see that actionable information."
Unlike PCR, which seeks to amplify targets of interest, CRISPRclean uses CRISPR-Cas guide RNAs (gRNA) to attach to sequences that aren't of interest and deplete them from a given sample. Then, the users are only left with the sequence or sequences that they want to study, or they're able to search for unknown signals that had been previously obscured by all the background noise.
"When you deplete, you're depleting the stuff you already know," said JumpCode Founder and Chief Technology Officer Keith Brown. "That enables unbiased discovery of everything else."
According to the company, the technology can be packaged into both research-use and clinical-use kits. Clinical kits for infectious disease could be used to remove human DNA for clearer signals of viral, bacterial, or fungal pathogen DNA. Oncology kits would increase the sensitivity of mutation detection by removing wild-type alleles from cancer patients' samples.
"I can take a saliva sample from you, remove all of the human molecules, and you're left with a single test for every viral, bacterial, and fungal pathogen in one shot," Brown said. "A global pandemic seems to be a good time to [use] that type of technology. Depending on what you remove in one test, you can get the virus, you can get the co-infection, and you can get the human immune response to determine which people are asymptomatic or are going to require more acute hospital care, all available on day one."
The clinical kits could also be used to improve genetic disease diagnosis, he added. While diagnosing genetic diseases can be made more accurate through whole-genome sequencing overlaid with RNA sequencing, the abundance of overrepresented genes, such as housekeeping genes, can make it very difficult to determine which genes are actually associated with the disease.
"But remove those, everything else comes up, and you dramatically increase your diagnostic yield of genome sequencing," Brown said.
On the research side, kits could be used for single-cell analysis by removing common transcripts in single-cell assays, thereby increasing their sensitivity to detect rare transcripts. They could also be used on whole-genome and whole-exome libraries by removing uninformative or repetitive sequences.
"We've identified a set of 100 or so genes that are overexpressed generally consistently across every sample type, so they contribute no material value information-wise but they soak up a lot of your sequencing reads," Hakak said. "Remove that, and now you ought to be able to get the improved sensitivity in that assay."
JumpCode isn't the first company to use CRISPR for the purposes of making the needle easier to find inside the haystack. In December 2018, Exact Sciences scientific cofounder and former chief technology officer Anthony Shuber launched a startup firm called Genetics Research, which uses CRISPR-based negative enrichment technology to make sample prep easier. The concept is to use CRISPR nucleases to protect a region of interest on a strand of DNA and then degrade the surrounding material, until the target fragments are all that are left to be sequenced and analyzed.
"Protect the needle, and then burn the haystack down," is how Shuber described it at the time.
The advantage of JumpCode's approach, according to Hakak, is that it seamlessly integrates into customers' existing workflows. The off-the-shelf kits require only a fairly short protocol, no new instrumentation, and can be done either pre-library prep or post-library prep, depending on the user's preference.
"And there's a value in doing it post-library prep, including being able to multiplex the samples instead of having to do this on every single sample," he added. "So, a 10x [Genomics] customer can create their standard single-cell library, and prior to going onto an Illumina sequencer, they perform this simple process to clean up that library and remove the content they don't want."
Further, the kit includes a tube of gRNAs that are designed to bind to target sequences, depending on the user's application. "We've developed a very robust pipeline that allows us to design these [guides] in very quick iterations," Hakak said. "And so, product iterations can ultimately be fairly quick — all we have to do is switch out the one tube, and you have a whole new application."
Today, the company has three kits available, he noted. There's a human ribonuclease depletion kit, a pan-bacterial ribonuclease depletion kit, which allows for the depletion of more than 200 different bacterial species, and a meta-transcriptomic depletion kit, which is a combination of the previous two.
The meta-transcriptomic kit, while useful for research purposes, would be the solution that JumpCode would look to use to create a universal infectious disease assay, Hakak said. The firm is also developing kits for whole-genome and single-cell applications.
"Two-thirds of the genome is generally repetitive information. Once you sequence it, you can't really map it… and we generally don't even use that information," Hakak said. "We can deplete that before we go on the sequencer, saving 70 percent of the cost of sequencing."
Importantly, he added, the kits require very low amounts of sample input. According to Hakak, the company has data showing the performance of its kits on single cells. And whereas most other library prep technologies require inputs in the range of about 10 nanograms, JumpCode's kits can work with inputs of about 100 picograms.
"That opens up your ability to process different types of sample that you previously couldn't," he added. "You're now able to see what you previously couldn't in a very simple, cost-effective way."
JumpCode also has "a dozen technologies" in its intellectual property portfolio making their way through its development pipeline, Brown said.
The company currently sells each of its kits for about $2,800. However, Brown noted, each kit can perform three reactions, and each reaction can handle up to 96 samples, so is a user maximizes the use of each kit, the cost per sample goes down to $10. It's "an order of magnitude reduction in per-sample cost" along with the "added benefits of smaller sample inputs and easier workflows," he said. And whereas a normal 96-sample RNA-seq study might take a week and more than 1,000 plastic pipette tips to process by doing the depletion work sample-by-sample, he added, a JumpCode kit can process all 96 samples "in a few hours with less than 10 pipette tips."