Scientists at the European Molecular Biology Laboratory (EMBL) in Heidelberg have created an AI-powered tool named MAGIC to identify cells with early chromosomal abnormalities linked to cancer. This system automates the detection of micronuclei, small DNA-containing structures that signal potential cancer development. The technology verifies a theory proposed over a century ago by Theodor Boveri.
The MAGIC system, short for machine learning-assisted genomics and imaging convergence, combines automated microscopy, single-cell sequencing, and artificial intelligence to study chromosomal errors in cells. Developed by researchers in the Korbel Group at EMBL Heidelberg, it addresses long-standing challenges in observing rare cellular defects that contribute to cancer.
Chromosomal abnormalities, such as changes in chromosome number or structure, are key drivers of aggressive cancers, associated with patient mortality, metastasis, recurrence, chemotherapy resistance, and rapid tumor growth. Jan Korbel, senior scientist at EMBL and senior author of the study published in Nature, explained: "We wanted to understand what determines the likelihood that cells undergo such chromosomal alterations, and what's the rate at which such abnormalities arise when a still normal cell divides."
The idea that irregular chromosomes play a role in cancer dates back to the early 20th century, when Theodor Boveri observed cells under a microscope and hypothesized their involvement. However, detecting these issues manually has been labor-intensive, as only a small fraction of cells show defects, and many are naturally eliminated.
MAGIC scans cell samples with an automated microscope, using a machine learning algorithm trained on labeled images to spot micronuclei—small compartments holding separated DNA fragments that heighten cancer risk. Upon detection, it tags the cell with a laser-activated photoconvertible dye for later isolation via flow cytometry, enabling genomic analysis.
Marco Cosenza, a research scientist in the Korbel Group, noted: "This project combined a lot of my interests in one. It involves genomics, microscopic imaging, and robotic automation. During the COVID-19-related lockdown in 2020, I could really spend some time on learning and applying AI computer vision technologies."
Testing on cultured normal human cells, the team found that over 10% of cell divisions result in spontaneous chromosomal abnormalities, nearly doubling when the tumor suppressor gene p53 is mutated. Factors like double-stranded DNA breaks also influence these errors. Collaborators included EMBL's Advanced Light Microscopy Facility, the Pepperkok Team, EMBL-EBI's Isidro Cortes-Ciriano group, and the German Cancer Research Centre's Andreas Kulozik team.
Korbel highlighted MAGIC's versatility: "As long as you have a feature that can be discriminated visually from a 'regular' cell, you can—thanks to AI—train the system to detect it." The tool processes nearly 100,000 cells in under a day, opening doors to broader biological research.
