ChatGPT at the service of Rosie: the tailor-made cancer vaccine for the dog born with Artificial Intelligence

The fate of Rosie seemed sealed: an aggressive mastocytoma, already subjected to multiple surgeries, had left her with very few clinical chances. However, her owner, entrepreneur and artificial intelligence consultant Paul Conyngham, did not give up.

Starting from November 2024, Conyngham embarked on a true bioinformatics analysis marathon led by ChatGPT. With a daily “deep dive” of two to three hours on AI, he mapped out the steps for an extraordinary endeavor: to convert the genetic signature of Rosie’s tumor into a custom mRNA vaccine.

During a procedure, veterinarians collected samples of tumor and healthy tissue, which were then sent to the Ramaciotti Centre for Genomics at UNSW, where the DNA was sequenced, producing about 350 gigabytes of raw data, at a cost of approximately $3,000. Once the results were obtained, the entrepreneur performed “variant calling”, comparing sick and healthy cells to identify the somatic mutations underlying the disease.

The analysis revealed the involvement of the c-KIT gene, commonly associated with canine mastocytomas. For the next step, Conyngham relied on AlphaFold, the software from DeepMind/Google that predicts the three-dimensional structure of mutated proteins. While requiring expert reading, the models allowed for the identification of neoantigens, the targets of the immune attack.

The principle is similar to that of Covid vaccines: to create an mRNA that temporarily instructs Rosie’s cells to synthesize fragments of the altered proteins, in order to train the immune system to recognize and selectively destroy the tumor cells expressing them. The uniqueness of the case lies in the double innovation: the accessible use of artificial intelligence tools to extract meaningful information, combined with the support of a high-profile academic infrastructure.

Indeed, Conyngham did not “cook” the drug at home: the crucial transition from the digital file to the purified molecule encapsulated in lipid nanoparticles took place in regulated environments and with dedicated equipment.

The UNSW RNA Institute, led by Páll Thordarson, transformed the provided data into a physical preparation.