AI-powered algorithm accelerates TCR identification to develop personalized immunotherapies
2024
German Cancer Research Center, Heidelberg, Germany
To accelerate the development of personalized T cell therapies, an antigen-independent classifier (predicTCR) is presented in the present study. PredicTCR is a machine learning algorithm that leverages high-performance single-TIL RNA sequencing techniques to identify reactive T cell receptors (TCRs) in tumor-infiltrating lymphocyte cultures (TILs) of various cancer types. The resected tumor tissue used for the study was provided by a male 62-year-old melanoma patient. After successfully identifying reactive TCRs in the donor's tumor samples, the researchers used the results and other sequence data to train the algorithm. The results show that predicTCR's predictions have significantly higher accuracy (up to 90%) and sensitivity than previous methods. Additionally, the combination of high-throughput TCR cloning and reactivity validation enables selection of prioritized TCR clonotypes in just a few days. In summary, the method proves to be an innovative and efficient approach that can help optimize the production of personalized immunotherapies by saving valuable time and continuously specifying and improving the search for reactive TCRs through the integrated machine learning system.
Prediction of tumor-reactive T cell receptors from scRNA-seq data for personalized T cell therapy
E. W. Green, M. Platten
Added on: 04-02-2024
[1] https://www.nature.com/articles/s41587-024-02161-y[2] https://www.bionity.com/en/news/1182979/machine-learning-classifier-accelerates-the-development-of-cellular-immunotherapies.html?utm_source=newsletter&utm_medium=email&utm_campaign=bionityde&WT.mc_id=ca0264
