Machine Learning Algorithm Identifies Tumour-Reactive T Cells

A novel machine-learning algorithm named PreGame has been developed to identify tumour-reactive gamma-delta (γδ) T cells using single-cell CITE sequencing data. This breakthrough, detailed in a publication in Nature on September 16, 2026, with the digital object identifier (doi) 10.1038/s41586-026-11055-9, offers a new method for pinpointing specific immune cells that can target and react to tumours. The research indicates that the expansion of this identified γδ T cell population can serve as a valuable biomarker for predicting and monitoring therapeutic response in patients.

Gamma-delta (γδ) T cells represent a distinct lineage of T lymphocytes that play a crucial role in innate and adaptive immunity. Unlike the more common alpha-beta (αβ) T cells, γδ T cells often recognize antigens in a non-MHC restricted manner, allowing them to respond rapidly to cellular stress and infection. Their unique properties make them promising candidates for cancer immunotherapy, as they can be engineered or expanded to target tumour cells directly. The development of PreGame addresses a significant challenge in this field: efficiently and accurately identifying which γδ T cells are specifically reactive to tumour antigens within complex biological samples.

Single-cell CITE sequencing (Cellular Indexing of Transcriptomes and Epitopes by sequencing) is a powerful technology that simultaneously measures both the transcriptome (RNA) and surface protein expression of individual cells. By integrating these two data types, researchers can gain a more comprehensive understanding of cell states and functions. PreGame leverages the rich information provided by CITE sequencing to distinguish tumour-reactive γδ T cells from other immune cells or non-reactive γδ T cells. The algorithm's ability to process and interpret this high-dimensional data is key to its success in identifying these specific cell populations.

The implication of PreGame's development extends to its potential as a biomarker. Monitoring the abundance and activity of tumour-reactive γδ T cells could provide clinicians with critical insights into how a patient's immune system is responding to cancer treatments. An increase in the population of these specific T cells might indicate that a therapy is effective, while a lack of expansion could suggest the need for alternative treatment strategies. This could lead to more personalized and effective cancer care, optimizing treatment plans based on individual immune profiles and responses. The research published in Nature lays the groundwork for further investigation into the clinical utility of PreGame and the broader application of γδ T cell-based diagnostics and therapeutics.