The first fruit born of 10x Genomics’ collaboration with AI company Anthropic has arrived.
10x announced on Nov. 2 the launch of Sentira Single-Cell, an “agentic analysis application” that allows researchers to explore and analyze single-cell sequencing data through the use of natural language inputs.
The software is powered by agentic AI using Anthropic’s Claude Opus — the second most powerful tier of models. Here’s how it works, at a high level:
Researchers can bring a scientific question or hypothesis to the application in natural language, or ask the agent to propose a hypothesis based on their data. The agent determines how to approach the question, develops an analysis plan for the researcher to review and refine, executes the analysis and returns results.
Here are some example prompts 10x provides in its tutorial for Sentira Single-Cell:
Based on the input datasets and experimental design, propose a novel hypothesis that is likely to be testable given the supplied data.
Pre-process and annotate relevant cell types in my dataset. What gene sets or pathways are enriched in each cell type?
What are the distinct tumor cell subpopulations in my sample?
And before you ask: yes, Sentira can generate UMAP plots for you.
“What stands out to me is how easily Sentira Single-Cell lets us ask questions of our data directly and get answers quickly,” Henry Long, scientific director at the Dana-Farber Cancer Institute, said in a statement. “The questions themselves aren’t new, but being able to ask them in natural language changes how fast we can get answers and move toward insights that matter for patients.”
For 10x, the software helps researchers cover the “last mile between generating data and answering the specific scientific question a researcher set out to explore,” said 10x CTO and Cofounder Michael Schnall-Levin.
In prior interviews, 10x CEO and Cofounder Serge Saxonov has told me that this is a huge bottleneck for the company to sell more reagents: researchers don’t want to generate more data when they haven’t been able to analyze what they’ve already collected.
Sentira Single-Cell uses consumption-based pricing and 10x is offering 100 promotional credits to start researchers off. The cost for agentic analysis can vary based on complexity, but 10x estimates that each session will cost approximately 12 credits. Data storage costs 0.02 credits per gigabyte per month and downloading data costs 0.1 credit per gigabyte.
That adds up to approximately 60 credits for a multi-sample experiment of fewer than 500,000 cells from start to finish: five agentic analysis sessions, 12 months of storage, and two full output downloads.
With the smallest “pack” of 100 10x Cloud credits for $120, that’s just under $75 per experiment. With the mega pack of 25,000 credits for $25,000, that’s $60.
10x was careful to distinguish its implementation from a simple Claude Code deployment. In an email, a 10x spokesperson noted that
“while Claude is currently the model we’ve chosen for Sentira’s natural language interface, Claude Code is not the agent harness. We’ve developed a custom harness which invokes Claude models to generate code, but importantly encodes 10x best practices specific to existing experience with 10x data.”
Other 10x Genomics News
10x and Stanford University have sued Waters in the Hamburg, Germany local division of the EU’s Unified Patent Court, alleging infringement of European Patent No. 3828285, titled “transposition into native chromatin for personal epigenomics,” a patent covering assay for transposase-accessible chromatin by sequencing (ATAC-seq) technology developed by researchers in William Greenleaf’s lab, including Jason Buenrostro, now at Harvard University.
There’s little documentation available, though the lawsuit is likely over the BD Rhapsody Single-Cell ATAC-seq assay, recently acquired by Waters.
The US version of the patent, licensed to 10x by Stanford, has already proved useful in the company’s patent fight against Parse Biosciences, now part of Qiagen, now owned by EQT. In 2025, Parse paid 10x $150,000 to settle allegations of patent infringement, though Parse claimed it wasn’t developing an ATAC-seq assay.
Separately, 10x announced it has appointed Ansuman Satpathy and Joseph Gilliam to its board of directors, effective Oct. 28.
Satpathy is an immunotherapy researcher at Stanford University and a venture partner at Wing Venture Capital. He has cofounded Immunai, Cartography Biosciences, Santa Ana Bio and Arpelos Biosciences. He was a scientific advisor to 10x prior to joining the board. He holds an MD-PhD from Washington University in St. Louis.
Gilliam is chief operating officer of Glaukos Corporation; he was also previously CFO. Prior to that, he was a healthcare investment banker at JP Morgan Chase. He is also on the board of Caris Life Sciences.
Other Genomics News
Illumina ‘Not About Just Selling Sequencers Anymore,’ CEO Says
Raw Reads is a series of refined, comprehensive transcripts of conversations on topics covered by Ion Genomics.
Illumina Q3 Revenue Grows 4 Percent, $37M Tariff Refund Headed to Customers
Some Illumina customers will soon see credits to their account for tariff charges passed onto them by the company from February 2025 to February 2026.
Illumina, Ultima Win Roles in Singaporean 450K-Person Genomics Study
Illumina and Ultima Genomics have each announced that they’ve been selected as sequencing providers for SG-INSIGHT, a 450,000-person cohort under Phase III of Singapore’s National Precision Medicine (NPM) program.
Tempus AI Q3 Revenue Jumps 23 Percent Driven by Oncology Testing
Tempus AI reported revenue growth of 23 percent in the third quarter, driven by data licensing oncology testing volume.
Guardant Health has appointed Kevin Sayer to its board of directors. Sayer is formerly CEO of continuous glucose monitoring technology company Dexcom and currently chairs its board of directors.
My Omics Reading List
Jackson et al., “Cancer-specific T cell receptor discovery using 150,000 clinical genomics samples.” Science Translational Medicine. Nov. 4, 2026.
Researchers from BioNTech and Tempus AI looked at bulk RNA sequencing data from approximately 150,000 patient tumor samples for T cell-receptor (TCR) sequences. They identified TCRs against therapeutically-relevant targets and identified TCR chains associated with well-known cancer antigens, including human papillomavirus, mutated KRAS, and CT83.
“Our results suggest that public anticancer TCRs are common and may represent a new source for the development of cancer therapeutics,” they wrote.
—
Zhu, Li, Chang, et al. “An open structure-guided RNA foundation model for generalizable prediction.” Nature. Nov. 4, 2026.
Researchers from the University of Science and Technology of China present structRFM, a foundation model for predicting RNA structure trained on 21 million RNA sequence–structure pairs.
“By jointly learning sequential and structural dependencies, structRFM produces discriminative sequence-level, nucleotide-level and pairwise features that support zero-shot inference and downstream structure and function prediction,” they wrote.
Elsewhere on Substack
The United States has added BGI and MGI Tech to its Uyghur Forced Labor Prevention Act (UFLPA) Entity List, “effectively blocking imports of their products unless importers can overcome a legal presumption that they were produced using forced labour,” writes Matt Marlowe at BridgeCross Bio.
BGI and MGI have rejected the allegations, he noted. Whether this actually impacts any business isn’t clear, as MGI is in the process of offloading Complete Genomics to Swiss Rockets.
Elsewhere on the Internet
In the US midterm elections, Democrats won a stunning number of races across the country at both the federal and state levels. Driving the rebuke of the Trump Administration included concerns about the cost of living, the war with Iran, and ICE agents killing Americans in the streets of Minneapolis; however, one of the top concerns was about AI, note Jacob Shamisian and Brent Griffiths at Business Insider. The so-called “blue tsunami”
was a backlash against data centers, the physical symbols of how artificial intelligence has been taking over the world. And it was a rebuke to Silicon Valley elites and their vision for the future.
They noted that an exit poll from NBC News found that three quarters of voters would oppose construction of a new data center in their area and more than half said AI and data centers were a “crucial” or “major” factor in their vote their congressional representative.
My prediction is that this means AI companies will go into overdrive to promote the “AI can cure cancer in X years” narrative. American politicians have already shown that cancer testing is one of the few things they’re willing to come together and pass legislation on and more generally, cancer doesn’t have many ardent supporters in the populace.
I’ll be covering it all here, on Ion Genomics.