Abstract
T cell is an indispensable component of the immune system and its multifaceted functions are shaped by the distinct T cell types and their various states. Although multiple computational models exist for predicting the abundance of diverse T cell types, tools for assessing their states to characterize their degree of resting, activation, and suppression are lacking. To address this gap, a robust and nuanced scoring tool called T cell state identifier (TCellSI) leveraging Mann-Whitney U statistics is established. The TCellSI methodology enables the evaluation of eight distinct T cell states-Quiescence, Regulating, Proliferation, Helper, Cytotoxicity, Progenitor exhaustion, Terminal exhaustion, and Senescence-from transcriptome data, providing T cell state scores (TCSS) for samples through specific marker gene sets and a compiled reference spectrum. Validated against sizeable pseudo-bulk and actual bulk RNA-seq data across a range of T cell types, TCellSI not only accurately characterizes T cell states but also surpasses existing well-discovered signatures in reflecting the nature of T cells. Significantly, the tool demonstrates predictive value in the immune environment, correlating T cell states with patient prognosis and responses to immunotherapy. For better utilization, the TCellSI is readily accessible through user-friendly R package and web server (https://guolab.wchscu.cn/TCellSI/). By offering insights into personalized cancer therapies, TCellSI has the potential to improve treatment outcomes and efficacy.