Scientists developed a new artificial intelligence model for predicting anti-cancer immunity

Oct, 2021 - By WMR

The data scientists and Researchers at the University of Texas have created an AI program that can determine which neoantigens are identified by the immune systems.

The pMTnet method may result in new techniques for predicting cancer diagnosis and immunotherapy response. Cancer cells' genomic mutations make them express several neoantigens on the surface. A few neoantigens are identified by T cells that seek indications of cancer and external substances, helping the immune system to kill cancer cells. Other neoantigens appear to be undetectable to T-cells, enabling cancers to expand unchecked.

Predicting the neoantigens identified by T-cells might assist researchers in developing customized cancer vaccines, improving T-cell treatments, and determining how patients would react to other forms of immunotherapies. However, there are so many unique neoantigens, and predicting which ones cause a T-cell reaction is slow, technically difficult, and expensive. With funds from the Cancer Prevention and Research Institutes of Texas, and the National Institute of Health, the team turned to AI in the discovery of a good alternative.

The researchers used data from recognized binding or nonbinding pairings of the three components - Neoantigens, proteins known as Major Histocompatibility Complex that traces neoantigen on cancer cell surfaces, and the T cell receptors are implicated for identifying the Neoantigen MHC proteins. The algorithm was then examined with a dataset gathered from 30 unique studies which had scientifically discovered nonbinding or binding neoantigen T-cell receptor pairings. This experiment demonstrated that their AI programs were highly accurate.

The researchers utilized this novel method to gain more understandings of neoantigens compiled in Cancer Genome Atlas, a database containing data of primary tumors. According to pMTnet, neoantigens usually trigger a greater immune response than tumor-related antigens. It also indicated which patients would respond better to immune checkpoint barrier treatments and had a higher chance of survival.