In short, the pipeline runs: alter genes across millions of cells, photograph them, let AI find the visual signatures, isolate the interesting cells, and analyze their protein composition.

What did the researchers find?

The team ran two proof-of-concept investigations reported in the Cell paper:

• Autophagy. In a genome-wide experiment, the researchers used SPARCS to study autophagy, the process by which cells recycle their own components. The AI analyzed microscopy images for changes caused by genetic alterations. "SPARCS identified a large proportion of the genes already known to regulate autophagosome formation, while also uncovering additional genes involved in the process," said first author Dr. Niklas Schmacke of LMU. Autophagosomes are the membrane sacs that cells use to wrap up material for recycling.

• Immune sensing. The researchers also examined STING, a key sensor of the innate immune system — the body's rapid, first-line defense. They discovered that the acidity of the Golgi apparatus, the cellular structure that packages and ships proteins, is important for STING's early transport within the cell. A protein called GPHR influences this pH level and, as a result, the subsequent activation of the immune sensor.

Why does isolating cells matter?

The recovery step is what distinguishes SPARCS from image-only screening methods. Many AI microscopy approaches can classify cells from pictures but cannot retrieve the physical cells for further study. SPARCS pulls out the cells of interest intact, which allowed the Munich-area team to run mass spectrometry on them and pin down exactly which proteins change when a given gene is disrupted.

That combination — visual screening plus molecular follow-up — means researchers can move from "this gene causes this look" to "this gene causes this look by altering these specific proteins." The study highlights this as a route to characterizing genetic effects precisely at the molecular level.

How solid are the results?

The published work is a proof of concept, and the authors present it that way. The autophagy screen validated SPARCS against genes already known to regulate autophagosome formation, which is a reasonable benchmark, and it turned up additional candidate genes. The STING findings on Golgi acidity and GPHR come from one biological system; whether the same mechanisms generalize to other pathways will require further study.

The researchers say the study demonstrates how AI, combined with scalable image-based genetic screening, opens new opportunities to systematically investigate genes and advance biological discovery. The 70 million cells analyzed in this initial study suggest the approach can operate at a scale that manual microscopy cannot match.

The paper is "SPARCS enables scalable recovery of complex image-based phenotypes for genetic screening," by Niklas A. Schmacke et al., Cell (2026), DOI: 10.1016/j.cell.2026.09.021.