Researchers at Yale University have developed a machine learning model called Immunostruct that can help scientists develop more personalized vaccines, including cancer vaccines. They described the tool as follows: nature machine intelligence We will also present the results of applying it to cancer and immunology data.
When a potential threat, such as a virus or tumor, occurs within our bodies, immune cells recognize peptides (essentially short proteins) on the surface of the invader and mount a defensive response. This small region with which the immune system interacts is known as an epitope.
Epitope-based vaccines are an emerging technology that contains specific peptides to elicit an immune response that precisely targets a specific disease. Ongoing research shows that these vaccines hold promise as immunotherapy for a variety of cancers, including melanoma, breast cancer, and glioblastoma. Researchers are also investigating whether these vaccines can fight new variants of the infection more effectively.
To develop these vaccines, scientists can use models that help predict which peptides are most likely to cause a strong immune response to a particular antigen. A limitation of many of these models, the researchers say, is that they treat peptides as one-dimensional amino acid sequences rather than three-dimensional active structures.
Now, researchers at Yale University have created a model that also incorporates the structural and biochemical properties of the peptide. In new research, They show that the multimodal model is more effective in identifying peptide candidates than previous models.
“Cancer is extremely heterogeneous, which often makes effective treatment very difficult,” says Kevin B. Gibsian, MD, PhD student at Yale University and co-senior author of the study. “We built a deep learning model that integrates more information than previously combined to improve the identification of vaccine targets that stimulate the immune system against people’s tumors. This allows for more effective and less toxic treatments.”
The Immunostruct model is available open source via GitHub.
