Researchers at the Korea Advanced Institute of Science and Technology (KAIST) have developed DeepNeo, an artificial intelligence (AI) model to predict neoantigens that can be used in personalized therapeutic cancer vaccines.

Neoantigens are protein fragments from mutations in cancer cells that can elicit an immune response and ideal targets for cancer vaccine development.

Multinational pharmaceutical companies, Moderna and BioNTech, are conducting clinical trials with large pharmaceutical companies for neoantigen cancer vaccines.

According to the KAIST team, led by Professor Choi Jung-kyoon, for a cancer vaccine to be effective, the mutations must not only bind to the major histocompatibility complex (MHC), but also the conjugates must trigger a T-cell immune response. However, that is not possible with current technology, it said.

The team noted that companies are conducting cancer vaccine clinical trials without knowing if the conjugates can stimulate an immune response. As a solution, it built a new conceptual deep learning model to solve these problems. It confirmed that it is possible to discover novel antigens with excellent immune and anti-cancer responses through various big data analyses.

Using the model, the team has also built a web service in collaboration with Penta Medix, a local company specializing in digital therapy. It expects companies to use the website to develop anti-cancer vaccines that effectively induce T-cell responses.

“As the mRNA platform has been validated through Covid-19 vaccines, we hope our website will help in the commercialization of mRNA-based cancer vaccines,” Professor Choi said.

Penta Medix CEO Cho Dae-yeon said, “We will accelerate the commercialization of personalized cancer vaccines using the platform.”

Researchers wanting to use the AI platform can visit the website -- https://deepneo.net.