And it may be the last chance science-minded regulators have to put some reins on Galleri before it gets out into a world where Grail has used its money and influence to tilt the market in its favor.

Alex Dickinson, a former Illumina executive who has built a large following on LinkedIn, in part due to his sharp critiques of Grail and Galleri, joined me to break the situation down.

Grail has spent years building up Galleri as a “multi-cancer early detection” test, but as Dickinson pointed out, “if you look at Grail’s [FDA approval] application, it doesn’t quantify either ‘early’ or ‘multi.’” The FDA advisory committee zeroed in on exactly that gap: what does “early” mean, and should Grail be allowed to use that word?

What the FDA decides to allow Grail to put on the test label and say in its advertisements may be as consequential as the approval itself. The decision will set precedent for Abbott/Exact Sciences’ Cancerguard test, Natera’s new test and Breakthrough device designation announced this morning, and others developing competing blood-based tests to promise to detect many cancers early.

Other key questions we try to address:

What do Galleri’s capabilities and shortcomings mean for the future of this kind of test?

How might the FDA regulate them in the future?

What impacts will they have on the US healthcare system?

You can also listen on Apple Podcasts and Spotify

Subscribe now

Show Notes

Alex Dickinson’ LinkedIn Feed

OpenOnco

Gilbert Welch STAT Op-ed

AdCom Vote Explanation Segment Transcript:

My story for GenomeWeb on Grail disclosing that the NHS-Galleri trial not meet its primary endpoint (Paywall)

David Cameron NHS/Grail/Illumina

https://www.bbc.com/news/business-58146567

BMJ investigation into Grail’s NHS-Galleri trial: “Galleri promises to detect multiple cancers—but new evidence casts doubt on this much hyped blood test”

Grail’s $15 million in lobbying:

Grail participation in RFK Jr.-linked event, first reported by Bloomberg (paywall) and also in the Los Angeles Times.

NHS Oncologists’ Letter to The Lancet: “GRAIL-Galleri: why the special treatment?”

Dan Swerdlow (Radiologist, Georgetown University) quote from AdCom meeting on the burden on the healthcare system for follow-up testing:

Natera said in an Oct. 9 statement that in granting the breakthrough status, the FDA assessed data from 12 cancer types and considered the test’s sensitivity, specificity, and performance on determining cancer signal origin. However, the company did not disclose the underlying results, the test’s name, how many cancer types it is designed to detect, or a development or launch timeline.

“Too many cancers are diagnosed after they’ve progressed, simply because limited screening options exist,” Alexey Aleshin, Natera’s chief medical officer and general manager of oncology and early cancer detection, suggesting that the designation reflected “the rigor and promise of our MCED program.”

As I reported earlier today, new commitments of approximately $800 million have poured in: the Energy Department is promising $500 million over five years towards lab work, AI modeling, and computation while Google DeepMind, Isomorphic Labs, and Meta are committing approximately $300 million towards the Virtual Biology Initiative (VBI.)

With $500 million already in the pot from Biohub, which launched the VBI, and data resources from NIH that Biohub said represent the end product of $500 million in previously issued federal funding, the resources going into this project are approaching levels likely not seen since the Human Genome Project.

The addition of US federal agencies adds major weight to the project, which had already signed on numerous privately-funded research institutes and academic consortia. But this time Biohub, itself a privately-funded organization (with direct ties to Meta CEO Mark Zuckerberg) appears to be driving the train. Whether everybody continues to sit together nicely remains to be seen.

A Biohub spokesperson said that “resulting data is meant to be an open resource for the research community, with Biohub catalyzing the effort to bring organizations together to build the layer that lets datasets work in a unified fashion,” but whether models built with these data will have similarly open access isn’t clear.

Building a virtual cell is a big goal with uncertain timelines and investment requirements. I will be following this story very closely with an eye towards history and how these kinds of Big Science projects can go right or wrong.

Ion Genomics News

Listen this Friday as I go deep into the FDA advisory committee meeting on Grail’s Galleri cancer test with Alex Dickinson, a former Illumina executive who has built a huge following on LinkedIn with his frank takes on the sequencing and cancer testing industries. We’ll discuss how money, power, and politics are clashing with data in the campaign to create a new kind of test: one that may or may not be able to call itself “early detection.” Find it on Apple, Spotify, or www.iongenomics.bio.

Subscribe now

Other Genomics News

Stanford University’s Karl Deisseroth has won a share of the 2026 Nobel Prize in physiology or medicine for his work on optogenetics, the technique that uses light to switch individual neurons on and off. Deisseroth shares the award, announced Oct. 5, with German researchers Peter Hegemann of Humboldt University and Georg Nagel of the University of Würzburg.

Deisseroth’s lab also produced the STARmap 3D spatial transcriptomics method, developed with Xiao Wang, now at the Broad Institute, while she studied under him. Deisseroth cofounded Stellaromics, which is commercializing STARmap.

Also of note: Feng Zhang of the Broad Institute and MIT, known for pioneering research on CRISPR-Cas9 genome editing in human cells, worked in Deisseroth’s lab at Stanford early in his career.

In the wake of the NHS-Galleri trial, clinical studies for blood-based cancer screening can use late-stage incidence as a surrogate endpoint for mortality, however, “late stage” should be clearly defined for a given cancer type, according to “The Peachtree Consensus” statement produced by a panel of experts convened by the American Cancer Society. As reported by Forest Ray of GenomeWeb, they published their recommendations on early cancer detection and multi-cancer detection trial design Sept. 23 in Cancer. Other important factors to consider are the “timing of screening intervals, the number of screening rounds, and follow‐ups after the last screening round,” Ray wrote.

My Omics Reading List

Cranley, J., Kanemaru, K., Bayraktar, S. et al. “Dynamics of human cardiogenesis and its disruption in trisomy 21.” Nature. Oct. 7, 2026.

Single-cell and spatial experiments helped researchers led by University of Cambridge’s Sarah Teichmann, one of the leaders of the Human Cell Atlas, define 21 distinct tissue niches in the developing heart.

“Comparative profiling revealed that hearts with trisomy 21 are depleted in compact cardiomyocytes and exhibit increased apoptosis relative to euploid hearts,” they wrote, a finding that was validated in experiments using heart cells derived from induced pluripotent stem cells. “These early developmental perturbations may contribute to the increased risk of congenital heart disease associated with Down syndrome,” they noted.

Elsewhere on Substack

Nava Whiteford teases unearthed details about Tilde Bio a “stealth(ish) nanopore company.” I was going to do my own dive based on this breadcrumb but ran out of time. I still might, but in the meantime, check out his work (paywall.)

ASeq Newsletter

Tilde Bio

A friendly Discorder contacted me about Tilde Bio (FL110) a stealth(ish) nanopore company backed by Flagship Pioneering…

Read more

5 days ago · 2 likes

The first of two projects plans to use 10x’s Flex Apex single-cell sequencing and Xenium and Atera spatial platforms to study tumor samples from patients with advanced bladder cancer treated with antibody-drug conjugates, immune checkpoint inhibitors, and other therapies. The goal is to correlate drug target expression and tumor microenvironment features with treatment response and clinical outcomes, then explore biomarkers that could one day help inform therapy selection.

The other project will use the Flex Apex platform to investigate residual disease detection in acute myeloid leukemia (AML). The organizations noted that monitoring for residual leukemic cells after treatment, which guides care, currently relies on a “patchwork” of separate technologies. 10x and Fred Hutch plan to investigate whether single-cell analysis can improve on current gold-standard methods.

Lawrence Fong, scientific director of Fred Hutch’s Integrated Immunotherapy Research Center and leader of the collaboration, said in a statement that the project could help characterize the cellular and molecular responses to cancer treatments and guide treatment decisions.

Additionally, 10x and Fred Hutch plan to build a framework for future clinical reporting, defining how single-cell and spatial features could support treatment planning in patients with cancer.

Intismeran autogene (V940/mRNA-4157) therapy, one of a class of therapeutics sometimes referred to as personalized cancer vaccines, is being evaluated in combination with Merck’s Keytruda (pembrolizumab) in patients with completely resected melanoma and other cancer types.

Under the agreement, announced Oct. 7, Tempus will manage the collection and transfer of the tumor tissue and blood samples for next-generation sequencing to identify neoantigens unique to the tumor. Subject to applicable regulatory approvals, Tempus will also provide NGS services to support the therapy’s design and manufacturing process.

Financial terms were not disclosed. The collaboration builds on joint efforts that began last year, according to Tempus.

“Bringing an individualized therapy like intismeran autogene from clinical development toward potential commercialization requires a seamless, highly reliable diagnostic network,” Said Francis, Moderna’s chief business officer and general manager of individualized neoantigen therapies, said in a statement. “Working with Tempus allows us to leverage their commercial footprint and advanced sequencing capabilities as we work to deliver individualized neoantigen therapies to patients at scale.”

On Oct. 7, Biohub, the organization behind the Virtual Biology Initiative (VBI) to create data for AI models, announced that it was partnering with the DoE and NIH. DoE will invest $500 million over five years to support laboratory measurements, AI modeling, and computation. NIH will contribute existing data resources and will work with Biohub to standardize datasets for use in training AI models.

Additionally, Google DeepMind, Meta, and UK-based AI-first drug discovery company Isomorphic labs are collectively contributing $300 million to the VBI.

In a statement, Biohub called it “the largest coordinated commitment to generating AI-ready biological data to date.” The result is intended to be an open data resource for researchers, a single point of access with a degree of standardization and uniformity.

“An accurate predictive model of biology could dramatically accelerate scientific discovery by enabling scientists to perform experiments digitally,” Biohub Head of Science Alex Rives said in a statement. “Because of this potential, the creation of a virtual cell is one of the most important challenges for the next era of science. It will require coordinated data generation efforts at a national and international scale, which is why these partners are coming together.”

Biohub is the main philanthropic outlet for Facebook cofounder and Meta CEO Mark Zuckerberg. In April, Biohub launched the VBI with $500 million and partnerships to create data with the Allen Institute, Arc Institute, Broad Institute, and Wellcome Sanger Institute, as well as consortia including the Human Cell Atlas, Billion Cells Project, and the Human Protein Atlas. Nvidia, which makes graphics processing units used in AI computation, has also signed on to provide computing resources, AI development tools, and technical expertise.

The partnerships align Biohub’s VBI with DoE and NIH’s respective programs to push the pace of scientific discovery using AI.

DOE funding will be used for fundamental cell research — data collection, AI analytics, measurement and imaging, modeling, and computation — drawing on supercomputing, imaging, and other laboratory resources from the department.

The Energy Department’s computing resources and facilities at the Joint Genome Institute will combine with Biohub’s expertise to “[set] a new standard for open science that will accelerate discoveries in both medicine and biotechnology,” Darío Gil, DoE’s Under Secretary for Science, said in a statement.

NIH’s contributions include national biomedical repositories catalogued by NIH’s National Library of Medicine and the National Center for Biotechnology Information, as well as NIH Common Fund programs that are already developing coordinated biological atlases, shared data standards, and AI-ready biomedical datasets.

In a statement, Biohub claimed the NIH data resources represent the output of approximately $500 million in previously allocated federal funding.

Whether Google DeepMind, Meta, and Isomorphic Labs are providing cash or in-kind contributions isn’t clear. Isomorphic labs is a spinout from Google DeepMind.

Researchers behind the COSMOS-MEL01 study analyzed 280 longitudinal surveillance samples from 48 patients with clinical stage II-III melanoma treated with surgery. During surveillance, Guardant Reveal detected 81 percent of distant recurrences and showed 100 percent specificity in samples from patients who remained recurrence-free. Circulating tumor DNA (ctDNA) detection preceded clinical identification of recurrence by a median of 70.5 days, and by up to 273 days in some patients.

The researchers published the results of the study Oct. 5 in Cancer Research Communications.

“Our findings suggest that serial ctDNA testing may provide a dynamic view of recurrence risk and potentially complement routine imaging, particularly in identifying patients at increased risk of distant relapse,” Kenjiro Namikawa, chief of the Department of Dermatologic Oncology at National Cancer Center Hospital in Tokyo and first author of the study, said in a statement. “Further studies will be important to determine whether and how ctDNA results could be incorporated into postoperative treatment and surveillance decisions.”

In a statement, Guardant said MRD detection about four weeks after surgery was independently associated with an approximately 12-fold increased risk of recurrence and a more than 30-fold increased risk of death compared to patients who tested MRD-negative at the same timepoint.

COSMOS-MEL01 was a prospective, multicenter evaluation of MRD testing in resected stage II-III melanoma. COSMOS (COnquer Solid Malignancies by blOod Screening) is a Guardant program evaluating MRD detection with Reveal across six cancer types.

Guardant Reveal is currently marketed for MRD detection in early-stage colorectal, breast and lung cancers.

“COSMOS-MEL01 extends the growing body of evidence supporting Guardant Reveal into melanoma and demonstrates the power of epigenomic MRD detection across a diverse set of melanoma subtypes,” Guardant Co-CEO and Cofounder Helmy Eltoukhy said in a statement. “The ability to identify patients at very high risk shortly after surgery, and to detect molecular recurrence before it appears clinically, could help inform more personalized surveillance and clinical trial strategies.”

The AFT-70 NAVIGATE trial, announced Oct. 5, pairs Genentech’s giredestrant, a selective estrogen receptor degrader, with longitudinal monitoring using Natera’s Signatera Genome MRD test.

AFT is sponsoring and leading the study, with collaboration and funding from both Natera and Genentech. Financial terms of the collaboration were not disclosed.

The study is designed to test whether patients can maintain outcomes when CDK4/6 inhibitors are used more selectively, rather than giving all patients prolonged combination therapy.

“We know that patients with ER-positive, HER2-negative breast cancer do not all face the same risk of recurrence, yet adjuvant treatment decisions are still largely based on clinical and pathological features measured at diagnosis,” Komal Jhaveri, the study chair and an oncologist at Memorial Sloan Kettering Cancer Center, said in a statement. “The trial will determine whether MRD-guided monitoring can distinguish patients most likely to benefit from adding a CDK4/6 inhibitor from those in whom it can be safely omitted. It’s an important step forward using biology—not just baseline risk—to guide adjuvant therapy.”

AFT expects to enroll more than 2,000 patients with stage II-III ER-positive/HER2-negative breast cancer who are eligible for endocrine therapy, at about 200 sites in the US and internationally.

The control arm will receive standard-of-care endocrine therapy plus upfront CDK4/6 inhibition. The investigational arm will receive giredestrant alone, with a CDK4/6 inhibitor added only if Signatera becomes positive. The primary endpoint is non-inferiority in four-year distant recurrence-free survival. The partners did not say when enrollment will begin or when results are expected.

The study is the latest de-escalation trial evaluating the use of MRD to help spare patients with solid tumors extra treatment without harming their survival odds. Trials using a similar approach in colorectal cancer include the DYNAMIC and VEGA.

Under the terms of the agreement, announced Oct. 6, the companies will generate data from full-length RNAs that TychoBio will use to train models to design RNA therapies. TychoBio will use PacBio’s Kinnex long-read RNA sequencing kits and HiFi sequencing on more than 10,000 samples. Financial and other terms of the deal were not disclosed.

TychBio plans to build large-scale datasets to link design of steric blocking antisense oligonucleotides to their effects across the transcriptome and plans to expand the approach to silencing RNAs and other RNA-targeting compounds. The results will be used as training data for models designed to predict which compounds are most likely to work, how long their effects may last, and where unintended effects may occur.

“AI models are only as good as the data they are trained on. To effectively identify new rare disease treatments, we needed the highest quality biological data, which can only be generated using PacBio HiFi sequencing technology,” TychoBio Founder and CEO Felix Raimundo said in a statement. “HiFi long-read sequencing lets us see across full-length transcripts, including changes in isoforms and splicing of how the transcriptome is affected in rare diseases and which treatments could modify disease processes.”

The Illumina MOU, also signed by representatives of the University of Latvia, Riga Stradiņš University, Riga East University Hospital, Pauls Stradiņš Clinical University Hospital and Children’s Clinical University Hospital, establishes a “framework for bringing modern genomics closer to patients in Latvia,” NIRI said in a statement. “This includes strengthening Latvia’s genomic data and biobanking infrastructure, developing local sequencing and bioinformatics expertise, and fostering closer links between researchers, industry, and healthcare.”

Similarly, the Element MOU addresses cooperation and collaboration in genomics, bioinformatics, and clinical research.

“The next era of precision health will be led by nations with the vision to connect biology, data, technology, and healthcare — and to build the pathways that turn scientific insight into better health for their citizens,” Illumina CEO Jacob Thaysen said in a Sept. 29 LinkedIn post about the MOU. “Together, we aim to advance genomics research, accelerate precision medicine, and improve outcomes for patients.”

The MOUs were part of a Latvian government trade mission to California, led by President Edgars Rinkēvičs. Generally, MOUs are non-binding agreements that signal intent to cooperate.

The agreements help establish credibility for the NIRI, formally created earlier this year from the merger of the Latvian Institute of Organic Synthesis and the Latvian Biomedical Research and Study Centre. NIRI controls the Genome Database of the Latvian Population and the national Genome Centre, as well as a network of university hospitals and researchers.

Latvia, one of the small Baltic states formed after the dissolution of the USSR, may also be seeking to establish its own identity in the world of genomics. Its neighbor to the north, Estonia, has already created a national biobank and population sequencing program linked to digital health records. Looking south, Lithuania’s Vilnius University has yielded important work on molecular technologies including CRISPR/Cas9 genome editing and single-cell analysis via spinout Atrandi Biosciences.

NIRI’s heritage includes drug discovery, potentially offering a differentiated way to capitalize on its national genome program. In the October 2025 announcement of its formation NIRI said it would pursue advances in “molecular medicine, biotechnology, and structural biology” leading to “development of novel medical solutions, and the commercialization of research results.”

For companies like Illumina and Element, NIRI projects itself as a potential strategic partner in its domestic industry and an additional entrypoint into the European market.

Illumina’s MOU builds on existing relationships with Latvian institutions and “Latvia’s growing capabilities and national ambitions in genomics and precision medicine,” an Illumina spokesperson said in an email. The company declined to disclose additional details about the MOU.

Element noted that NIRI already has one of its Aviti systems, but declined to offer additional details of the MOU.

Ion Genomics News

Adding to the global focus of this week’s newsletter, my latest podcast was an interview with BridgeCross Bio’s Matt Marlowe on China’s burgeoning biotech industry, including sequencing technology companies.

We produced a few short videos for social media such as this one about the competitive landscape there, which Matt also posted to LinkedIn.

Let me know what you think of it and if you’d like to see more pulled from future (or past) episodes.

Other Genomics News

Anthropic has claimed that its Claude AI program has discovered “a novel enzyme system that is associated with an array of DNA repeats, a pattern reminiscent of CRISPR.”

However, Denmark-based researcher Mario Rodriguez Mestre told Carl Zimmer of The New York Times (paywall) that he had identified the same system and used Claude as part of his research, which has remained unpublished. The story is reminiscent of another AI-powered “discovery,” where mathematician Tristan Buckmaster alleged that OpenAI cribbed his ideas about how to solve a famous math problem.

As part of the enzyme discovery announcement, Anthropic revealed that it operates a wet lab in the Bay Area to validate some of Claude’s biological predictions.

Imagenomix is a spinout of New York University and boasts cancer researcher and pathologist Matija Snuderl as one of its cofounders.

Caris Life Sciences announced Sept. 29 an enhanced version of its Caris Assure liquid biopsy test. It integrates whole-exome and whole-transcriptome sequencing with whole-blood analysis. The test includes validation that tumor fraction correlates with risk of progression and treatment efficacy in patients who receive repeated testing, as well as validated Human Leukocyte Antigen (HLA) reporting, expanded pharmacogenomic reporting, and an enhanced clinical report.

Senticell has won a $400,000 Small Business Technology Transfer Phase I grant from the National Cancer Institute to fund research into a red blood-cell-bound tumor DNA liquid biopsy test for early detection of non-small-cell lung cancer (NSCLC.)

“Our preliminary work suggests red blood cells may retain tumor DNA longer and in greater abundance than plasma,” Senticell Cofounder Nilam Mangalmurti, also a researcher at the University of Pennsylvania Perelman School of Medicine, said in a statement. “This grant lets us rigorously test that biology, first in animal models and then in a cohort of 100 early-stage NSCLC patients, to determine whether it can become a reliable diagnostic tool.”

EpiSign has launched EpiSign Metric 5-base, methylation analysis software that will expand the company’s capabilities beyond microarrays to sequencing-based DNA methylation assays.

The software can analyze methylation information generated using Illumina 5-base, Pacific Bioscience’s HiFi sequencing, and Oxford Nanopore Technologies’ sequencing through the same classification framework used for array-based analysis. The software provides automated analysis across more than 300 episignature disorders.

My Omics Reading List/Elsewhere on Substack

On his Substack Ground Truths, Eric Topol of Scripps Research dives into two papers that propose a new model of aging: rather than simply accumulating damage, cells may also lose their identities as specific cell types, through erosion of epigenetic programs dictating that identity.

Ground Truths

Loss of Cell Identity Drives Human Aging

Read more

10 days ago · 229 likes · Eric Topol

One of the papers, a review titled “Written in development and lost in ageing: the grammar of cellular identity” was published Sept. 30 in Nature.

The funding will continue to support efforts to decode the biological drivers of type 2 diabetes and obesity and identify new therapeutic targets by combining large-scale genomic data generation, artificial intelligence, and cell models.

Specifically, the center will expand two research programs: the Human Gene Regulation Map, which studies how genetic variants affect disease-causing genetic circuits and programs; and Flagship Disease Projects, which examine disease biology through protein-coding genes, molecular pathways, and cell functions.

“Cardiometabolic diseases including diabetes and obesity are the leading cause of death globally. Yet we still don’t understand enough about the biological mechanisms that drive them,” Mads Thomsen, CEO of the Novo Nordisk Foundation, said in a statement. “By uniting Denmark’s world-class expertise in metabolic and population research with the Broad’s unparalleled genomic and computational capabilities, we’re already seeing advances that might not otherwise have been possible.”

Since launching in 2021 with an initial $47.5 million grant, the center has grown into a collaboration spanning 18 research groups and more than 120 scientists. According to the Broad Institute, it has produced resources for cellular and genomic aspects of metabolic disease as well as technologies for establishing function of variants, datasets linking genetics to cell biology, and tools for genetic discovery across diverse populations.

Over the next five years, researchers plan to generate hundreds of millions of detailed cell profiles, including from fat cells and brain cells grown from stem cells, to train AI models that predict how DNA changes disrupt cellular health.

The commercial launch positions the test as an alternative for patients who decline or do not complete a preferred stool-based (e.g. Cologuard Plus, acquired by Abbott via Exact Sciences) or a direct visual screening test, such as endoscopy.

“We still see too many eligible patients who remain unscreened. Expanding the ways patients can participate in screening gives clinicians another opportunity to address those barriers,” Folasade May, a gastroenterologist and director of quality improvement in gastroenterology at UCLA Health, said in a statement. “For someone who has declined or does not complete preferred approaches, the option of a blood draw may be what helps them finally take action.”

SimpleScreen CRC uses circulating cell-free DNA from blood to detect signals associated with colorectal cancer in adults age 45 and older at average risk for the disease. The test was developed and manufactured by Freenome and is exclusively commercialized by Abbott.

In June, Freenome announced SimpleScreen CRC had met all primary and secondary endpoints in a clinical validation study.

Under a partnership announced Sept. 29, the companies aim to combine AI-based screening with targeted spatial mutation confirmation, with the goal of expanding access to mutation testing. Initial work will focus on a workflow for clinically actionable oncology mutations such as KRAS, EGFR and AKT, while the companies build a joint roadmap for future validation studies and broader scientific collaboration.

“Imagenomix detects actionable mutations directly from the H&E slide every patient already has. Pairing that with Haga Bio’s in situ confirmation in the same tissue gives us a path to bring precise mutation testing to far more patients, faster,” Heather Gustafson, senior director of program strategy at Imagenomix, said in a statement.

The collaboration is supported by Vinnova, Sweden’s Innovation Agency, through an initiative to strengthen international collaborations. Financial and other details were not disclosed.

New York-based Imagenomix predicts clinically actionable mutations from routine H&E pathology images. Founded in 2023, it is a spinout from New York University, cofounded by cancer researcher Matija Snuderl, among others.

Stockholm-based Haga Bio develops spatial assays that detect specific mutations in formalin-fixed, paraffin-embedded tissue samples. In June, the company announced a $2.2 million in seed financing round.

On Sept. 29, 10x announced the opening of an early-access program for Sentira, a cloud-based platform that offers GPU-accelerated data analysis pipelines, as well as visualization and collaboration tools it says can greatly accelerate research timelines.

The product employs hardware-accelerated algorithms pairing Nvidia’s graphics processing units (GPUs) with the rapids-singlecell analysis suite co-developed by Nvidia and the open-source scverse consortium, which maintains widely used single-cell analysis tools.

Under the early-access program, the spatial version of Sentira will be available for customers using 10x’s Xenium and Atera spatial instruments, with a planned launch as a paid product in early 2027. Pricing will be based on “consumption” with enterprise subscription options, 10x said, but did not disclose any additional pricing details.

How the pricing is set could be key to adoption, according to Nick Banovich, a researcher at the Translational Genomics Research Institute (TGen) who also runs a single-cell and spatial core lab.

“GPU-enabled analysis is critical,” he said. “For these big spatial datasets I’d say it becomes necessary. It’s not really tractable to have some of the front-end analyses take 10 to 20 days, especially if they are things you need to iterate on, like [cell] clustering.”

Beyond Sentira’s ability to quicken spatial data analysis, its collaboration capabilities are “a huge advantage we see,” 10x Cofounder and Chief Technology Officer Michael Schnall-Levin told Ion Genomics. Sentira also places its capabilities further downstream than 10x’s free Cell Ranger and Space Ranger suites have in the past. (According to 10x, Atera and Xenium generate “cell feature matrices” as part of the instrument run, obviating the need for Cell and Space Ranger.) For example, Sentira will have its own method for cell annotations, something that was only available as a service from 10x before.

Schnall-Levin said there are ”no current plans” to sunset any of 10x’s major software products.

The Sentira roadmap also includes support for single-cell sequencing data as well as agentic AI analysis and more capabilities further down the line from data generation; however, 10x did not provide a specific timeline for those additions.

“The idea is to cover the entire portfolio,” Schnall-Levin said.

Is Grail’s Galleri test safe and effective? Do its risks outweigh its potential benefits?

The blood-based cancer detection test passed all three votes before the ten members of the Molecular and Clinical Genetics Devices Panel of the FDA’s Medical Devices Advisory Committee on Sept. 23.

The panel voted unanimously in favor of Galleri on the question of safety, but was split on the other two questions. The vote was closest, 6-5 in favor, on whether Galleri was effective at its proposed use of detecting multiple cancers early, and 7-2 in favor (with one abstention) on whether the rewards outweighed the risks presented by the test.

After hours of testimony and discussion, the ten panelists explained their reasoning behind each vote. Their comments reveal concerns about the use of the word “early” and broad questions about what other tests like Galleri will mean for the US healthcare system, as well as hopes for what these tests can do for detecting cancers for which there are not established screening methods.

These votes will guide FDA as it mulls approval for Galleri, though the agency is not bound by the results.

What follows is a refined, corrected, and annotated transcript of the video of the meeting posted publicly on YouTube; however, some comments were not attributable to a specific speaker and are noted that way.

Question One: Safety (10-0)

Unknown speaker

[...]We’re going to vote on all three and then we go around the room and discuss each one individually. [...] Question one: Is there a reasonable assurance that Galleri is safe for patients who meet the criteria specified in the proposed indication? [...]

Question two: Is there reasonable assurance that Galleri is effective for use in patients who meet the criteria specified in the proposed indication? [...]

Question three: Do the benefits of Galleri outweigh the risks for use in patients who meet the criteria specified in the proposed indication? [...]

I’m going to start with Dr. Ghosh and just state for your name how you voted and what influenced your vote. It could be short because you’re going to have to do it three times.

Aditya Ghosh (Internist physician, Mayo Clinic)

Aditya Ghosh Mayo Clinic. Yes. Uh, reasonable assurance that it’s safe. I think especially when put with uh when accompanied by uh adequate directions and warning against unsafe use. I think we’ve had extensive conversation too in terms of what that looks like and so really important in making sure that it’s appropriately rolled out and um that everybody’s aware of how it is intended to be used.

Kelly Hunt (Breast cancer surgeon, University of Texas MD Anderson Cancer Center)

Kelly Hunt. Yeah, I voted yes. I think as was stated previously, it’s safe. I believe it’s safe based on the data that we’ve seen and that along as an adjunct to appropriate cancer age appropriate cancer screening is safe.

Victor Van Berkel (Thoracic surgeon, University of Louisville)

I’m Dr. Van Berkel. Uh I also voted yes. Um I think this obviously the safety question covers a lot of territory. I think false reassurance is the largest issue leading to either ignoring symptoms or uh other screening guidelines. I think within the statement of safety, the part when accompanied by adequate directions and warnings carries a lot of weight. Um, and in that context, I think that it’s safe.

I do struggle a little bit with what to do with when it is obvious that the target audience will misinterpret the results. Just for the record, we didn’t talk about this, but this is evidenced by some of the public comments. There were a couple of physicians in the public comments who said, and you know, specifically something along the lines that this test not only finds cancer, but it rules it out, which this test clearly does not do. And there were two or three physicians, I don’t know if you guys read through all 180 of the comments there, but there were a couple people who definitely said that. And so, it’s not just patients that are going to misinterpret this, but physicians that are going to misinterpret this, but that’s not on the company, that’s on us. But with that in mind, I still think that they have done what they can as far as that goes.

Deborah Armstrong (Oncologist, Johns Hopkins University)

Um, Deb Armstrong, um, I voted yes. Um, I do think, uh, it’s safe. The biggest concerns, as we’ve talked about, um, are people being falsely reassured and not getting, um, recommended screening. Um, and then the very small number of people who have a false positive and the exposure they get to testing. Um uh but if I think if people are well informed, I think that’s an acceptable um that’s acceptable.

Jason Dominitz (Gastroenterologist, University of Washington and the VA)

Jason Dominitz, I also voted yes. Um echoing prior comments, you know, my biggest concern’s that the direct safety of the test itself is pretty clear minimum risk of the direct direct uh side effects of the test, but indirect uh risks are real uh in terms of false reassurance and false positives. And I think that may change with age. And I think it will be really important for us to think about an upper bound to minimize the risk to those who would not benefit from the test. And so labeling should consider things along the lines of, you know, intended for people for whom screening is appropriate. And there’s going to be a lot of people for whom screening is not appropriate. So we need to keep that in mind.

Charity Morgan (Biostatician, University of Alabama at Birmingham)

Charity Morgan. Like Dr. Armstrong. I was swayed by the low rate of false positives and the low number of invasive procedures in the false positives. Um the other possible risk that I was worried about was the impact on uh recommended screenings in people who are getting the tests, but we don’t have any evidence at this time that that is being effective. So what I’m left with in front of us today is evidence that we do see that this is a safe test.

Dan Makarov (Urologist, New York University)

Dan Makarov of NYU. I voted yes uh that I felt it was safe uh for reasons previously stated, but to summarize uh um uh low risk of false positive um and with uh you know appropriate uh guidance uh uh with restrictions of how it’s meant to be used and then also with uh with hope with my hope that there’s postmarket surveillance.

Lori Dodd (Statistician, National Institute of Allergy and Infectious Disease)

Lori Dodd and I voted yes for many of the reasons stated so I won’t go through them again.

Stan Lipkowitz (Senior Investigator, National Cancer Institute)

Stan Lipkowitz I voted yes and again for the same reasons. [...]

Dan Swerdlow (Radiologist, Georgetown University)

Dr. Swerdlow. Yeah, I voted yes. For the record, for the record, I voted yes. Um, for all the reasons uh previously mentioned, I do think there needs to be robust education of the practitioners who going to order this test so they really understand what it means uh more than just the label, but they probably should do an online course or something that gives them a good uh explanation of the data such as it is.

The section is available to all subscribers. It’s free, so sign up now:

Subscribe now

Question Three: Benefits vs Risk (7-2)

Unidentified speaker

On question three, the panel voted seven yes, two nos, and one abstain that the benefits of Galleri outweigh the risk for use in patients who meet the criteria specified in the proposed indication.

Dominitz

Jason Dominitz. Well, you heard me talk about my heartburn over the answer to question two, but once I made a determination of yes for that and say if it was just a logical conclusion, but again, I have significant concerns and would like to see true outcome data, not necessarily mortality. I think that’s too high of a bar for a test like this. I mean, we don’t hold, you know, Cologuard to that standard. We randomized trial colonoscopy just recently, you know, hasn’t shown enough mortality benefit from colonoscopy, but we have other data to support that. But, you know, I think we have to be careful not to set too high of a bar, but I do think we need more outcome data for this test.

Morgan

Uh, Charity Morgan. Um, I did vote yes to this one. This one was a hard call as well. Not as hard as the last one. Um, honestly, you know, post approval studies could change my mind on this one, but um, you know, that with the absence of data to the contrary, I think what we have in front of us says that the benefits um, outweigh the risk at this point.

Makarov

I voted yes. I thought it was safe. I thought it was effective and the breakdown of risk risk benefit I thought was favorable one again, being driven a lot by cancers that don’t have um uh current screening protocols available and with the hope that there’s robust postmarket surveillance.

Dodd

Lori Dodd I voted no. You might wonder how I could vote yes on the other two and no on this one. Um but that’s because I took this to be a broader set of questions and and rather than reading the literal um label and the literal definitions of safety and effectiveness um you know do the benefits outweigh the risk? And in order to really address that, I felt like we needed to know more about clinical outcomes um and you know are these cancers that need to be treated? Some of them likely do, but how many of them don’t need to be treated? So, I would like to see more of those data before voting yes on this question.

Lipkowitz

So, I don’t know if this test is effective and you know, so that’s why I voted no, but that means that I don’t know if the risk and benefit where the risk and benefit come down. So, I abstain because I really just don’t know the answer to that question. and

Swerdlow

Yeah. Um so like Lori’s reasoning, I don’t know because we don’t have that data and I suppose someday it will become available but not right now. Um but I’m very concerned about the cost and implementation of this to the whole health care system. Um for things I discussed on the last question about number of providers, I see a lot of patients and providers twisting in the wind, not sharing, not sure what these results are actually going to mean. Uh as a contrast, lung cancer screening where everybody has to go through uh talk to doctors, get counsel, get understanding what these positive lung cancer screening tests mean. I know there’s a lot of backup for patients there. And in this case, I just don’t see that backup. And I see a lot of patients who are going to struggle to get their workups done because it’s not going to be convenient for them or it’s not going to get paid for by insurance or it’s simply not going to be cost effective to a system that’s already struggling.

Ghosh

Dr. Ghosh, I voted yes that the benefits of Galleri outweigh the risks. I think as the discussions and as everyone we’ve heard from on the panel you know there are certain things that you know Dr. Dominitz and I had brought brought up initially that at one point it’s you we wonder is that outside the scope of the conversation that we’re having today but I think it’s really important to say there’s a lot that goes into benefits and risks that are within scope and survival is definitely on there and the overall you know ability of this to um you know the overall ability of uh Grail to ensure that they have appropriate support uh for teams going forward uh when they’re rolling out this test is really important. Um, I think those were all kind of very valid points. I think I was weighing those against the risk of also not doing this. And when we say that there are cancers that we don’t currently have screening for, um, well, I don’t know if that’s true anymore.

You know, when we, you know, the cancer signal of origin, the things that we’ve seen and the fact that things like head and neck cancers, hematologic malignancies, um, some of the pancreatic malignancies, some of these that we know can be devastating to patients, the fact that there’s a blood test that could possibly capture them. And, you know, in terms of weighing that risk benefit ratio, I certainly think there is a lot of work that has to be done in order for us to definitively say stuff going forward. But when I weigh the risk and benefit of a test that could possibly give us something where we don’t currently have something um that’s just too much of a risk in my mind. So that’s why I weighed yes.

Hunt

Kelly Hunt I voted yes for a lot of the same reasons that have been articulated by Dr. Ghosh. And I also feel like the information that was provided about the algorithms that are utilized to help uh navigate the process of false positives and then also retesting for the negatives I think is very helpful in utilizing this test and then hearing from physicians who do utilize it. I think we learned the good and the bad part of that. So, I think the FDA can take a lot of that information to help in the labeling about how it should be utilized and what the yes means or what a positive test means and what a negative test means.

Van Berkel

Victor Van Berkel and I voted yes. I feel like Dr. Lipkowitz and I were thinking about this very much the same way and I seriously considered abstaining as well and then I chickened out. I guess I don’t know. Um, I feel like the only logically appropriate thing for me to say after saying yes and then no is that I’m not sure whether or not the risks are outweighed by the benefits. Uh I suppose the reason that I hedge towards yes is because I think that in the years ahead it probably will be proven to have the benefits outweigh the risks specifically when it comes to the tests that the cancers that we don’t have good tests for. Um I think even a marginally successful test if we can find uh more stage one or more early stage cancers in a in a in a population that we can’t screen for currently then that will end up having benefit. But I don’t know that that’s been proven yet. So perhaps my yes is a hopeful one rather than an accurate one.

Armstrong

I also voted yes. I don’t have a whole lot to add. I did think that the risk and my biggest concern would be the false um false positive and the rate is really quite low and I think that’s very reassuring to me. So, I think when you um um I think the benefits the potential benefits for this I think it but it is really potential at this point in time. So, um outweigh the risk. So, other than that, I don’t really have anything else to add.

Unidentified Speaker

I think uh at this point in time I’m supposed to summarize the vote. We had on the question of safety we had 10 yes and zero no. On the question of um effective we had six yes and four no. And then on the question of uh do the benefits outweigh the risks we had seven yes two no and one abstain. Um anything else anybody wants to contribute at this point in time? If not, um I’d like to thank everybody here. We take a lot of time out of our busy lives to do this. Um like to thank the FDA folks who’ve uh really put a huge amount of work in here and the sponsor um and the folks uh who um uh for the uh open public hearing both in person and online um many of which really were quite moving and remind us of the humanity that’s behind some of this. I think we’re ready to adjourn.

My podcast guest this week, Matthew Marlowe, visited Illumina’s China operations as part of a two-week tour around the country to meet with and discover companies based there. He runs BridgeCross Bio, a digital publication seeking to provide insight into China’s biotech industry for a Western audience, alongside his boutique life science marketing firm, Genapse.

While some of high-level restrictions placed on Illumina have been removed, it appears that selling sequencers remains challenging. Still, the company should have hope, Marlowe suggested.

“There’s strong brand recognition there and I think respect for the technology and that’s not just gonna disappear,” he said, adding that Western biopharma’s interest in China is only increasing. “That’s a big, presumably addressable segment for Illumina.”

In the larger picture, China’s burgeoning biotech and genomics markets make for the ultimate compare and contrast.

Like the US, it has regional hotspots, intense competition, and a big company whose success has helped seed a second wave of startups, many of whom are now breaking through.

However, the clinical system is both vast and vastly different, marketing approaches can seem stuck in the past, and success abroad — rather than at home — may end up being the difference between making it or not.

We discuss some of the new tech he’ll be watching and what it means for companies based in the US and Europe. Also, Marlowe share’s his favorite roast goose restaurant in Hong Kong.

Subscribe now

You can also listen to the episode on Apple Podcasts and Spotify

Show Notes

BridgeCross Bio’s China companies map

BridgeCross Bio’s Substack

BridgeCross Bio

Your Bridge to China Biotech

By Matt Marlowe

Genemind’s recent financing round:

LinkZill microarray synthesizer

WuXi AppTec’s successful legal challenge of inclusion on the US DoD 1260H list

BridgeCross Bio

WuXi Won in Court. Can It Undo the Chill?

WuXi AppTec has won an important first round in its legal battle with the US Department of Defense…

Read more

2 months ago · Matt Marlowe

Oxford Nanopore’s legal fight with MGI Tech

Matt’s email

On Sept. 23, the Molecular and Clinical Genetics Devices Panel of the FDA’s Medical Devices Advisory Committee voted 6-4 that there is reasonable assurance that Galleri is effective for its proposed use of detecting multiple cancers “early.”

The six panelists who voted “yes” agreed the test clearly detects cancers, though several said they still wanted post-market studies and label language reinforcing that patients continue with guideline-recommended screening tests, such as colonoscopy for colorectal cancer. The four “no” votes reflected different concerns.

“The question comes down to what do you call clinical effectiveness?” said Stan Lipkowitz of the National Cancer Institute. “As an oncologist, I generally look at that as outcomes, and we don’t have that data.”

Another “no” vote, Daniel Swerdlow, a radiologist at Georgetown University, also questioned whether there was any demonstration of effectiveness and also raised concerns about the impact Galleri would have on demand for follow-up testing.

“There simply aren’t enough physicians to do this,” he said. “We already don’t have enough trained endoscopists in this country to do the routine screening colonoscopy that needs to be done in this country” and there is a “massive shortage of radiologists,” as well.

“It’s one thing for one company to start this, but when they all start coming online, they’re not going to get read in any timely fashion because there just simply aren’t enough bodies to go around further,” he said.

Charity Morgan, a biostatistics expert at the University of Alabama at Birmingham, voted “no” specifically because Grail’s proposed label still used the word “early,” and suggested she would change her vote to yes if the label were changed.

Another, Lordi Dodd of National Institute for Allergy and Infectious Disease, said she voted “yes” on effectiveness “assuming the label is changed to remove the ‘early’ part... that’s why I voted yes, but I have a lot of concerns about its broader clinical impact long term.”

The same panel was unanimous, 10-0, on the test’s safety, and voted 7-2, with one abstention, that Galleri’s benefits outweigh its risks.

The labeling question ran through much of the meeting. The panelists ultimately coalesced around dropping “early” in favor of simply “detection,” and agreed the test should be described as detecting a “cancer signal” shared across tumor types, verbiage taken from Grail’s own proposal for its test label.

“The panel agreed on calling this a test for a ‘cancer signal’ but also want to be clear that it does not detect ‘all cancer’ in order to mitigate risks that patients forgo other screenings,” Guggenheim Securities Analyst Subbu Nambi wrote in a Sept. 23 note to investors. “While the label may remove the “early detection” language in favor of simply “detection”, we don’t think this will matter in clinical practice,” she added.

Grail stock again jumped on news of the vote, up 15 percent in Sept. 24 trading on the Nasdaq at $125.21.

“Today’s vote reinforces the strength of Galleri’s clinical evidence,” Grail CEO Josh Ofman said in a statement.

Galleri, which the FDA designated a Breakthrough Device in 2018, is supported by data from the PATHFINDER 2 and NHS-Galleri studies. Grail submitted its premarket approval application on Jan. 29, 2026, and the FDA is expected to issue a final decision in the coming months. The FDA is not bound by the committee’s recommendation.

The panel meeting and subsequent vote is a hugely important step for Grail’s Galleri early-cancer detection test as well as for the entire multi-cancer early detection (MCED) market. Galleri is the first and so far most thoroughly-studied test, but Abbott’s Exact Science and others have similar tests in development.

Following the failure of the UK NHS’s Galleri trial to meet its primary endpoint of a reduction in late-stage cancers, disclosed earlier this year and published Sept. 22 in the New England Journal of Medicine, FDA approval is a chance for Grail to get positive news out there again: and an FDA stamp would definitely be positive.

Investors seem to think that at least some form of FDA approval is likely. While this is not the finish line for Galleri, it is hugely important. The “Medicare Multi-Cancer Early Detection Screening Coverage Act,” which was signed into law earlier this year, opens the door for Medicare to cover Galleri and other MCED tests, provided they’re approved by the FDA.

Without the new law, Grail would probably have had to continue generating evidence to support a US Preventive Services Task Force recommendation. Instead Grail dumped money into Washington’s K Street to help it blaze a shortcut.

According to mandatory disclosures, since 2020 Grail has spent more than $15 million lobbying congress to provide medicare coverage for MCED tests, including millions on the “Medicare Multi-Cancer Early Detection Screening Coverage Act” by name. That money went to over a dozen different lobbying firms.

As noted by genomics commentator Alex Dickinson on LinkedIn, the committee seems keen to push Grail by zeroing in on what “early detection” means and if the test could be approved simply to detect multiple cancers without that modifier.

Also at stake is how this FDA decision sets precedent for later MCED tests. Public comments from liquid biopsy experts led by Johns Hopkins University researcher and Exact Science Cofounder Bert Vogelstein noted that:

As an accompanying op-ed in NEJM suggested, we are only a step closer to a world in which cancers can truly be detected early. What happens with this FDA AdCom meeting will be another step: in which direction, one cannot say.

Ion Genomics News

Tune in to the podcast channel on Friday to hear about the Chinese genomics industry from Matt Marlowe, a life science marketing professional who also runs a Substack about the sector. He is fresh off a multi-week trip around the country doing site visits and taking meetings. Conveniently, Chinese President Xi Jinping is in the US this week to talk trade, AI, and other topics with the US government.

Subscribe now

Other Genomics News

White House Eyes Takeover of NIH Funding Decisions

As first reported by the Washington Post, the White House and NIH Director have discussed a proposed executive order that would give the White House unprecedented power over NIH grant making.

Science’s Joceyln Kaiser (soft paywall) notes that under a proposed executive order, a new board co-chaired by the director of the White House Office of Management and Budget Russell Vought would be able to veto any grants. What could go wrong?

Illumina ascended to the S&P 500 on Sept. 21. The index of US stocks is a key barometer of the US stock market. Shares of the DNA sequencing giant are up 92 percent over the last six months.

Twist Bioscience will supply antibody characterization data services for Eli Lilly’s TuneLab, an artificial intelligence-based drug discovery platform. Twist’s shares are up even more than Illumina’s: 275 percent over the the same period.

As MRD testing volumes rise for companies like Natera, Guardant Health, and others, so too are insurance denials, reports Alexandra Byrne of GenomeWeb (paywall.)

The US Department of Health and Human Services filled eight seats on the US Preventive Services Task Force. The seats had been vacant for months and the full panel hasn’t met in over a year.

Many of the new members are specialists and not primary care physicians, a fact that drew criticism from American Medical Association President Willie Underwood:

“The composition of this new slate represents a significant departure from the Task Force’s traditional membership,” he said in a statement. “We believe maintaining a strong primary care voice must remain central to its work.”

Protein analysis company Nuclera today announced the appointment of Carl Raimond as CEO effective immediately. Raimond was formerly president and chief commercial officer of Olink, now part of Thermo Fisher Scientific. Cofounder and former CEO Michael Chen will transition to the role of chief scientific and business officer.

My Omics Reading List

Lee, D., Koutrouli, M., Masse, N.Y. et al. “Single-cell atlas of transcriptomic vulnerability across brain disorders.” Nature Sept. 23, 2026.

The centerpiece of a nine-paper package published across Nature’s family of journals, this study presents data on a single-cell transcriptome atlas of more than 6 million nuclei from 1,494 unique donors, including patients with eight different conditions across dementia and mental illnesses, as well as “neurotypical” controls.

“By analyzing millions of cells across multiple brain disorders, we found that many diseases share common molecular pathways while also exhibiting distinct cellular signatures,” Donghoon Lee, first author of the paper and a researcher at the Icahn School of Medicine at Mount Sinai, said in a statement. “These findings provide a framework for understanding why different neurological and psychiatric disorders converge on similar biological processes while maintaining distinct disease-specific features.”

–

Yoon et al. “A noncontiguous code for RNA-guided DNA recognition at the origin of CRISPR-Cas.” Science. Sept. 17, 2026.

Researchers led by Jennifer Doudna of the University of California, Berkeley, describe what appears to be an “ancient progenitor of CRISPR systems.”

Dubbed viral interference programmable repeats (VIPR), these may evolved in viruses to help them fight off other viruses – weapons that bacteria later turned against their originators.

Also check out the long-form press release from the Innovative Genomics Institute’s Andy Murdock.

Elsewhere on Substack

Nava Whiteford, writing on his Substack, said that:

[ChatGPT has] done some quite interesting nanopore data analysis and given an Illumina MiSeq imageset built a complete image analysis and basecalling pipeline.

I previously wrote a pipeline to do this. It took me about a year, ChatGPT did it in about 3 hours.

Check out his article for his full list of prescriptions for data documentation and accessibility.

ASeq Newsletter

AI First Analysis For Startups

I’ve been thinking a lot about AI recently. I’ve mostly been an AI skeptic…

Read more

18 days ago · 2 likes

Elsewhere on the Internet

The Love Hypothesis movie is finally out on Prime Video. Check back next week for my official review!

Also:

The approval, announced Sept. 22, supports using Signatera to guide treatment with Genentech’s atezolizumab (Tecentriq) in MIBC patients.

Natera expects to commercially launch Signatera for MIBC in Japan in the first half of 2027, following the PMDA’s approval of atezolizumab in MIBC, and pending final reimbursement and pricing decisions.

This is Signatera’s second approval in Japan. In June, PMDA approved it as a companion diagnostic in colorectal cancer. The US FDA has also approved Signatera as a companion diagnostic for the same MIBC and atezolizumab indication.

The approval was supported by data from IMvigor011, a randomized, double-blind Phase 3 trial, which included more than 20 participating sites in Japan.

According to Natera, MIBC affects approximately 7,500 people in Japan each year.

Under the agreement, announced Sept. 16, Twist will supply antibody characterization data services for TuneLab, including for AbLab, an antibody developability prediction model.

“The quality of an AI model’s output depends on the quality of the data used to train it,” Twist CEO and Cofounder Emily Leproust said in a statement. “Twist has the scale and high-throughput capabilities to generate consistent and reliable antibody characterization data these models need to be trained on.”

TuneLab gives customers access to AI drug discovery models trained on Lilly’s proprietary research data. Through the new agreement, TuneLab users will be able to order antibody services from Twist using TuneLab-preferred protocols to generate wet lab data.

In a statement, Twist said the arrangement also lets TuneLab users obtain preferred pricing while contributing the resulting data back to TuneLab for federated training. Other financial terms of the deal were not disclosed.

The deal is another positive indicator for Twist’s position in the AI-enabled drug discovery services and data market. Last month, the company’s shares spiked after Anthropic revealed it had used Twist’s wet lab services to verify its work on engineering proteins with AI. Earlier in August, Leproust told investors that she expects revenue related to AI-enabled drug discovery to approximately double from 2026 to 2027, to about $100 million.

Shares of Twist Bioscience are up 17 percent since Sept. 16 at $166.99 and are up 275 percent over the last six months.