10x Genomics launches AI agent for single-cell biology analysis

Bridging the coding gap

Biologists studying single-cell data often face a bottleneck when trying to interpret their results. While they may have samples from treated and untreated cells, they frequently need a bioinformatician to write specific code to compare the two groups. This technical hurdle is known as the 'last mile' problem in single-cell analysis.

10x Genomics addressed this issue by launching Sentira Single Cell on November 2. The cloud application allows researchers to skip the manual coding steps usually required by software packages like Seurat or Scanpy. Instead of writing scripts in R or Python, users can simply ask questions about their data in plain English.

By moving the analysis into a natural language interface, the company hopes to democratize data access. As Michael Schnall-Levin, chief technology officer at 10x Genomics, told R&D World, the goal is to let people who do not write code have more direct access to their own research data.

How the AI agent works

The core of the new application is an AI agent powered by Anthropic’s Claude. When a researcher inputs a question, the agent proposes a specific plan to find the answer. The user has the opportunity to review and revise this plan before the agent actually executes the analysis.

Once the plan is approved, the agent runs the analysis on the 10x Cloud platform. It produces the final results and also provides the underlying Python code used to generate them. This ensures that the process remains transparent and reproducible for the scientists involved in the study.

Each run of the agent generates an editable notebook. These notebooks serve as a permanent record of the methods and code used during the session. This allows researchers to keep track of their workflow while still benefiting from the automation provided by the AI agent.

Impact on research speed

The primary benefit of this new tool is the speed at which researchers can receive answers to their questions. Henry Long, a scientific director at Dana-Farber Cancer Institute, noted that while the questions themselves are not new, the ability to ask them in natural language changes the pace of discovery. This allows for a much faster transition from hypothesis to data interpretation.

The application is designed to handle complex queries, such as identifying which specific cell types responded most to a certain treatment. Because the analysis runs on the cloud, labs can pursue multiple different questions in parallel. This concurrency can significantly shorten the time spent on iterative testing.

To ensure the system works correctly, 10x tested the agent against datasets that had been analyzed by hand by a computational biologist. The team compared the AI's results to the known answers provided by the human expert. While the company has not yet published the specific accuracy figures, they plan to release the results and the framework soon.

Pricing and future tools

10x Genomics is offering Sentira Single Cell on a consumption-based pricing model. To encourage adoption, the company is providing promotional credits to help new users get started with the platform. This model allows labs to pay for the specific amount of analysis they perform rather than a flat subscription fee.

The company is also expanding its AI capabilities with a companion application called Sentira Spatial. This tool has been in free early access since September 29 for researchers using specific 10x instruments. It utilizes NVIDIA GPUs and a package developed by scverse to speed up analysis cycles from hours down to minutes.

While Sentira Single Cell is currently available, Sentira Spatial is expected to move to a paid launch in early 2027. These tools represent a broader shift toward agentic workflows in biology, where AI handles the heavy lifting of data processing. The goal is to let scientists focus on the science rather than the software engineering required to see it.