A century old puzzle in cancer biology is finally getting a powerful new ally. According to a new report from SciTechDaily, scientists at EMBL Heidelberg have built an AI driven system that can track the earliest genetic mistakes that push healthy cells toward becoming cancer. These tiny errors, known as chromosomal abnormalities, have long been suspected as one of the first steps that send normal cells down a dangerous path.
For more than 100 years, researchers have known that abnormal chromosome numbers and structures often appear in aggressive tumors. As far back as 1914, German biologist Theodor Boveri proposed that mismatched chromosomes might drive cancer formation itself. But studying these early defects has always been extremely difficult. Only a small number of cells show abnormalities at any given moment, and most either die or get eliminated before scientists can examine them.
Researchers in the Korbel Group at EMBL realized that solving this mystery would require a completely new approach. They built a fully automated system called MAGIC, short for machine learning assisted genomics and imaging convergence. MAGIC combines robotic microscopy, advanced computer vision, and single cell genome sequencing to capture and analyze rare cells as they begin to develop structural chromosome defects.
The system works like an automated game of laser tag. First, a microscope scans thousands of living cells. An AI algorithm trained on manually labeled images identifies cells that contain a micronucleus, a tiny bubble of DNA that often signals a cell is on the verge of developing new chromosomal abnormalities. When the AI finds one, it instructs the microscope to tag the cell with a photoconvertible dye using a precise laser flash. This allows researchers to isolate those specific cells later and analyze their genomes in detail.
What once required days of manual searching can now be done in hours. MAGIC can process nearly 100,000 cells in less than a day. Using the system, researchers discovered that more than 10 percent of normal cell divisions produce spontaneous chromosomal defects. When p53, a key cancer suppressor gene, is mutated, that rate nearly doubles.
The team also mapped how DNA breaks and other triggers contribute to chromosome instability. Because MAGIC can be trained to recognize any visible cellular feature, it may soon help unlock mysteries far beyond cancer.
Scientists say the technique opens a new window into the earliest moments when a healthy cell begins to turn dangerous, potentially changing how cancer is understood and eventually detected. The study has been published here.