In a recent LinkedIn post, Dr. Riam Shammaa, CEO of closely held Cell Technologies, and co-author of a recently published paper in Frontiers in Immunology, highlighted findings from an investigation into whether genetically engineered stem cells could identify potential cancer immunotherapy targets that conventional computational models fail to predict.
“We have spent years developing AI models to predict biology. But living cells already possess something that computational models struggle to replicate: the ability to process biological information within its actual cellular environment,” Dr. Shammaa says. “This is what I refer to as Genetic Intelligence.”
The paper, titled “Mapping the prediction-invisible fraction of an engineered immunopeptidome: a pilot analysis of immunoproteasome-reprogrammed mesenchymal stromal cells,” describes research conducted by Dr. Shammaa and his team, who modified the antigen-processing machinery of mesenchymal stem cells and exposed them to tumor-derived material. They then analyzed the peptides presented by these cells and compared them with predictions from established computational models. “The results were interesting,” Dr. Shammaa says.
The results revealed that of the 234 peptides experimentally identified, 34 (14.5%) were missed by three computational prediction outputs. This was approximately 20 times the prediction-miss rate observed in an independent reference dataset.
“These findings confirm our earlier published research that demonstrated that these gene-engineered cells are superior to dendritic cells in killing different types of cancer in vivo due to a larger and more immunogenic repertoire. Nevertheless, while the identified peptides still require functional validation, they raise an important question about how we approach drug discovery,” Dr. Shammaa says.
He explains that AI models learn from existing biological data. Living cells, on the other hand, actively process proteins, respond to their environment, and generate biological information that may not be captured by those models.
“Rather than trying to predict every biological interaction computationally, why not use living cells to identify what actually happens?”
Dr. Shammaa believes this is where genetic intelligence becomes particularly interesting, especially in cancer immunotherapy, where identifying previously overlooked antigens could open new avenues for therapeutic development.
“The objective is not to replace AI. It is to combine the biological processing capabilities of living cells with AI’s analytical capabilities. AI predicts biology. Genetic Intelligence lets biology reveal itself,” he concludes.
