Gastric cancer (GC) ranks fifth for cancer incidence and fourth for mortality globally.1 Clinical outcomes have varied among patients receiving similar treatments at the same stage, suggesting the current prognostic tools remain somewhat flawed.2,3 Single-cell analysis of GC data allowed us to dissect transcriptional programs underlying lymphocyte residency and exhaustion. Combined with tumor purity, we developed a novel classification system and data analysis following the workflow shown in Figure S1.