Scientists Find Hidden Cancer Drug Target AI Missed

Scientists just found a secret doorway that could lead to cancer treatments with fewer side effects, and they discovered it in a place cutting-edge artificial intelligence never thought to look.

Researchers at the Icahn School of Medicine at Mount Sinai identified a previously unknown binding site in PKMYT1, a protein that controls how cells grow and divide. When this process goes wrong, cancer can develop.

The discovery matters because most experimental cancer drugs target the same crowded spot on proteins called the ATP-binding site. Since many proteins share nearly identical ATP-binding sites, drugs often hit the wrong targets, causing unwanted side effects.

The team used AI tools like AlphaFold2 to predict the protein's structure, then followed up with laboratory experiments. That's when they found something unexpected: a completely hidden pocket where drugs could attach, offering a new route to more selective treatments.

"AI was very accurate when predicting known protein shapes, but it missed a completely unexpected binding pocket that we could only uncover experimentally," says Dr. Avner Schlessinger, who directs the AI Small Molecule Drug Discovery Center at Mount Sinai. The hidden site may provide a way to design cancer drugs that zero in on their target without affecting other proteins.

The researchers made another surprising discovery during their experiments. Tiny chemical changes to a molecule dramatically altered how and where it bound to the protein, revealing that these structures are far more flexible and dynamic than scientists previously understood.

The team tested their findings using X-ray crystallography, biochemical testing, and cellular studies. They then ran the results through additional AI systems including AlphaFold3 and Boltz-2 to see if current computational tools could predict the newly discovered binding behavior. They couldn't.

Why This Inspires

This breakthrough shows the irreplaceable value of human curiosity and hands-on experimentation, even as AI transforms drug discovery. The researchers didn't dismiss AI when it missed something important. Instead, they used it as a starting point, combined it with traditional laboratory work, and found something neither approach could achieve alone.

The compounds identified represent promising starting points for developing future therapies. Next, the team plans to create more potent drugs targeting this newly discovered site and search for similar hidden pockets in other cancer-related proteins.

They also hope to teach AI systems to recognize these hard-to-detect protein shapes, making future discoveries faster. The work demonstrates that the future of medicine isn't about choosing between human insight and artificial intelligence, but about combining both to achieve breakthroughs neither could reach alone.