Researchers pair AI cell screening with physical isolation in a 70-million-cell study

A new cell-screening method promises to connect what AI sees under a microscope with the genetic changes behind it. In a study published in Cell, researchers demonstrated SPARCS across 70 million cells, then isolated selected cells for molecular analysis. The evidence centers on two biological processes: cellular recycling and activation of an immune sensor.

The team brought together researchers led by Veit Hornung at LMU’s Gene Center, Matthias Mann at the Max Planck Institute of Biochemistry, and Fabian Theis at Helmholtz Munich. Their method combines AI, microscopy and genetic screening.

From a visual feature to a recoverable cell

Genetic screening links changes in genes to changes in cells. The study’s dataset description identifies a bottleneck: genome-scale screens are routine for simple cellular traits, but complex features visible in microscope images remain harder to screen. SPARCS addresses that problem by selecting individual cells based on their images and physically recovering them.

SPARCS stands for spatially resolved CRISPR screening. In the study, researchers ran genome-wide CRISPR knockout experiments, which disable genes, and used machine learning to analyze the resulting cell images. Automated laser microdissection then isolated selected cells in place, allowing researchers to connect an observed cellular effect to its underlying genetic alteration.

Recycling genes and an immune sensor

The first genome-wide proof-of-concept experiment examined autophagy, a cellular recycling process. First author Niklas Schmacke said SPARCS recovered a large proportion of genes already known to regulate autophagosome formation, while also identifying additional genes involved in the process. That test therefore included both recovery of established biology and identification of additional genetic contributors, rather than only a count of screened cells.

The second experiment focused on STING, a sensor in the innate immune system. Researchers found that acidity in the Golgi apparatus matters for STING’s early movement within the cell. The protein GPHR influences that acidity and, consequently, the sensor’s subsequent activation. The finding connects a change in the cell’s internal chemical environment to transport and activation of an immune-system component.

What the recovered cells revealed

Because the selected cells remained intact after isolation, the team could examine their protein composition using mass spectrometry. This gave the researchers a molecular readout of the genetic changes, alongside the visual features used to select cells. The protein analysis revealed disruption involving the endoplasmic reticulum and Golgi, according to the study’s associated dataset description.

The researchers also used computational perturbation modeling to nominate additional STING regulators. Those are proposed regulators, distinct from the reported GPHR finding. The associated proteomics dataset is listed in ProteomeXchange as PXD082653 and was announced on October 6, 2026, providing a public data record tied to the Cell study.