AI-assisted proteomics investigates rare liver disease

As the field of proteomics continues to advance, scientists are using new methods to answer longstanding questions about rare diseases.

One such disease is α-1-antitrypsin deficiency (AATD), a proteotoxic liver disease, in which misfolded and unfolded proteins cause cell death. In this case, a deletion in the gene for α-1-antitrypsin leads to a misfolded mutant, which accumulates and eventually results in cell death. But how these aggregates harm cells and result in downstream damage like scarring—which only some people with AATD develop—has remained a mystery that cutting-edge proteomics could help illuminate.

Researchers in Matthias Mann’s lab at the Max Planck Institute of Biochemistry used deep visual proteomics (DVP) to investigate the structural differences between cells that accumulate these misfolded proteins and those that don’t (Nature 2025, DOI: 10.1038/s41586-025-08885-4).

First described by the Mann lab, DVP starts with tissue samples and uses microscopy to select individual cells that are then cut out by a laser for analysis by mass spectrometry (Nat. Biotechnol. 2022, DOI: 10.1038/s41587-022-01302-5). In this case, artificial intelligence helped select which cells to analyze from liver biopsy samples.

“Essentially, what we can pinpoint with the human eye is now done by a machine,” says Florian Rosenberger, the first author on the new paper.

The team used its method to track how the molecules in diseased cells change over time and the disease’s progression. The researchers found that in cells from people with AATD, oxidative organelles called peroxisomes were induced much later in those who got sick than in those who didn’t get sick. The relationship between the peroxisome and the unfolded proteins is unknown. But Rosenberger says that early peroxisome induction could have a protective effect, which might in turn suggest new treatment options for AATD.