PacBio Plugs Partnerships, Accuracy at JP Morgan
By Allison Proffitt
January 14, 2022 | Closing out the genomic presentations at the JP Morgan Healthcare Conference, Christian Henry, president and CEO of Pacific Biosciences, had the dual challenges of presenting PacBio’s year and future strategies as well as answering claims and announcements other companies made. He persevered.
He reported growth and expansion in 2021, including doubling the number of global locations and year-over-year growth of 65%. The install base grew by 84%, Henry said, and SMRT cells shipped grew by 220%. The vast majority of PacBio customers are using the HiFi long reads that the company first announced in January 2019, he reported.
In outlining the market wins for PacBio, Henry highlighted the microbial genomics market that PacBio entered with its HiFiViral SARS-CoV-2 kits. “This product was conceived in the early part of Q1 as we saw the pandemic continuing to move… and continuing to see that there was a need to be able to do surveillance without the need to constantly be updating the primers associated with the kits.” The kit launched in November 2021, and already Henry reports that they’ve shipped over 16,000 samples to 20 customers.
The clinical utility of long read sequencing in rare and undiagnosed disease has been an area of focus for PacBio for a while, but Henry also emphasized the market opportunity in oncology. “We have an emerging opportunity in oncology, and we see this as an incredible opportunity for us to grow over the next several years,” he added.
PacBio announced several new partnerships and collaborations this week ahead of the event, and Henry and colleagues touched on all of those.
In a research collaboration with Google, PacBio will explore the use of Google’s genomic analysis, machine learning and algorithm development tools to further improve PacBio’s already highly accurate variant calls for HiFi sequencing runs, unlocking more insights from PacBio sequencing data. The collaboration builds on previous research from PacBio and Google. A recent publication on the work available on bioRxiv, Deep Consensus: Gap-Aware Sequence Transformers for Sequence Correction, yielded improvements in variant calling and suggested that Google’s DeepConsensus machine learning tool is capable of increasing the yield of 99.9 percent accurate HiFi reads by as much as another 27% per instrument run.
“We’re excited to work with them on this,” said Mark Van Oene, Chief Operating Officer at PacBio. “Traditionally Google’s been doing a lot of work with DeepVariant and looking at interpretation and sort of backend processing. This time we’re having them focus really on our raw base accuracy. Something as small as a 2% increase in raw base accuracy with our traces could lead to 20% in yield. The goal here is to really get them inside the system and looking at the raw bases that includes methyl-C. So how do we think about raw base accuracy which is going to help reduce CCS coverage required, increase yield overall, and just drive the output and the throughput and the performance of the system.”
Berry Genomics is collaborating with PacBio to bring HiFi long read sequencing technology to the Chinese clinical market. Under the agreement, PacBio will develop its first long read desktop sequencing platform with funding and guidance provided by Berry Genomics. When the agreed product requirements are met, Berry Genomics will purchase at least 50 systems for use and sale throughout China. The two companies’ collaboration includes development, regulatory authorization, manufacturing, and distribution. The desktop long read sequencing instrument will help Berry Genomics expand its footprint in the perinatal and carrier testing market. Berry Genomics will also sell the instrument and consumables in China. This will enable laboratories across China to access PacBio long read sequencing, in an easy-to-use, small footprint desktop format, at a lower capital investment than prior PacBio systems.
During the presentation, Henry emphasized that with the Berry Genomics collaboration, the high throughput system would be a top priority. “Those will drive us well below the $1,000 genome and really help us in our push of making PacBio HiFi long reads the clinical standard in the long run. We’re going to focus on that first, and then feather in the desktop system,” he said. But the eventual desktop system, he said, will help Berry run clinical panels and diversify their testing capabilities.
Finally, the company announced a collaboration with Genomics England to study PacBio’s technology for identifying genetic variants in unexplained rare disease cases. The 100,000 Genomes Project had been able to find actionable mutations in around 25% of patients with rare disease, Henry explained, quoting Parker Moss, Chief Ecosystem & Partnership Officer at Genomics England.
“For the 75% of families that have unexplained variation or unexplained disease, we’re going to do long read sequencing in a pilot,” Henry said. “If that pilot is successful, it’ll provide us with an opportunity to help even more families.
The read length debate is ongoing. PacBio made its own diversification acquisition earlier this year, acquiring Omniome in July for $800M for its short read technology. And for what it’s worth, Clive Brown, CTO at Oxford Nanopore, spent several opening minutes during his presentation at Oxford Nanopore’s virtual Community Meeting in December emphasizing his platform’s ability to produce short reads.
And so once the prepared presentation was over, it was unsurprising that Henry faced questions about Illumina’s announcements of both new chemistry and a long read technology earlier in the week.
Illumina’s long read technology—code-named Infinity—will deliver contiguous data up to 10kb in length with 90% less DNA input, Illumina CEO Francis deSouza said in his presentation.
PacBio’s Henry was not terribly impressed. “They did not give all of the details,” he said. “In one of the presentations, one of my counterparts said 10kb aren’t long reads. Fair enough.” He made some guesses at how the 10kb reads are generated (deSouza did not give details in Illumina’s presentation), and concluded that, “The applications are probably more in the bacteriology and some of those smaller genomes.” He raised issues of sample prep, accuracy, and cost, workflow, platform compatibility.
“We’ll see and learn as the rest of the community, but from our perspective, we feel very confident in the technology that we have and that we’re going maintain being the leader in long read sequencing. Period. Full stop.”
Illumina also announced new chemistry earlier in the week—again with a code name, Chemistry X, and scant details. “At least from the claims they made during their presentation, it’s going to be dramatically less accurate than [PacBio’s] sequencing-by-binding chemistry. And that’s really going to be the hallmark—it’s that accuracy. It matters an incredible amount when you’re doing deep counting applications,” Henry said.
He also highlighted that if Chemistry X requires new hardware—and deSouza’s presentation intimated that it could—then existing Illumina customers will be open to a platform change. “If it requires any new hardware then, quite frankly, it gives us an opportunity to get into each of those customers when the time is right with our SBB and our highly-accurate offerings,” he said. “Given at least the specs that they’ve shown so far, it doesn’t threaten the high differentiation of SBB chemistry.”