EMBL researchers developed MAGIC, an AI-powered platform that combines microscopy, machine learning and laser tagging to identify and isolate rare cells with micronuclei—early indicators of chromosomal errors that can lead to cancer. MAGIC can screen over 100,000 cells per day, allowing study of events that were previously too rare to capture. The team found that more than 10% of cell divisions produce chromosomal abnormalities, a rate that nearly doubles when the tumor suppressor p53 is suppressed. Developers say MAGIC can be retrained to detect many other visual cellular features, widening its applications in biology.
MAGIC: AI System Uncovers the Earliest Chromosomal Errors That Can Trigger Cancer
EMBL researchers developed MAGIC, an AI-powered platform that combines microscopy, machine learning and laser tagging to identify and isolate rare cells with micronuclei—early indicators of chromosomal errors that can lead to cancer. MAGIC can screen over 100,000 cells per day, allowing study of events that were previously too rare to capture. The team found that more than 10% of cell divisions produce chromosomal abnormalities, a rate that nearly doubles when the tumor suppressor p53 is suppressed. Developers say MAGIC can be retrained to detect many other visual cellular features, widening its applications in biology.
05:51 PM, 11/10/2025Science
AI-assisted MAGIC spots the earliest genetic mistakes that can lead to cancer
Researchers at the European Molecular Biology Laboratory (EMBL) have developed an AI-driven platform called Machine Learning-Assisted Genomics and Imaging Convergence (MAGIC) that helps detect the earliest chromosomal errors linked to cancer. By combining high-resolution microscopy, machine learning, and laser-based tagging, MAGIC lets scientists identify and isolate rare cells that carry genetic defects for in-depth analysis.
Cancer arises when cells divide uncontrollably, often because of mistakes in their genetic blueprint. Chromosomal abnormalities—numerical or structural defects in chromosomes—are a known source of aggressive tumours and are closely associated with metastasis, recurrence, chemotherapy resistance, and patient mortality.
"Chromosomal abnormalities are a main driver for particularly aggressive cancers, and they're highly linked to patient death, metastasis, recurrence, chemotherapy resistance, and fast tumor onset," says Jan Korbel, senior scientist at EMBL.
Detecting these abnormalities is challenging because only a tiny fraction of otherwise healthy cells exhibit them at a given time, and many such cells are eliminated by natural selection before they can be studied. MAGIC is built to overcome these obstacles by enabling high-throughput, targeted capture of the rare cells of interest.
How MAGIC works
1) Imaging: High-resolution time-lapse microscopy captures images of dividing cells. 2) AI detection: Machine learning models scan images for micronuclei—small, membrane-bound structures that contain fragments of chromosomal material and are strong predictors of future chromosomal abnormalities. 3) Laser tagging: Flagged cells are illuminated with a laser that activates a dye, creating a durable colour tag on those specific cells. 4) Isolation and analysis: Tagged cells are separated using techniques such as flow cytometry and subjected to genomic and molecular analyses.
Where manual inspection once limited researchers to a handful of cells, MAGIC can screen more than 100,000 cells per day, making it possible to study rare, early events in cell division at scale.
Key findings and broader potential
Using MAGIC, EMBL researchers reported that over 10% of cell divisions produce chromosomal abnormalities, and that this proportion nearly doubles when p53—a key regulator of cell-cycle control and tumor suppression—is suppressed. These results provide new, quantitative insight into how frequently chromosome-level errors arise during cell division and how cellular safeguards like p53 help limit them.
Beyond cancer research, the developers emphasize MAGIC's versatility. "As long as a feature can be visually distinguished from a typical cell, the AI can be trained to detect it," Korbel explains. This means MAGIC could be adapted to isolate cells with a variety of visual phenotypes, accelerating discoveries across many areas of biology.
Implications: MAGIC is not a diagnostic tool for patients yet, but it is a powerful research platform that can reveal the earliest steps of tumorigenesis and help scientists investigate mechanisms of genomic instability, drug resistance, and disease progression.