Our immune system relies on an enormous variety of T cell receptors (TCRs), the molecular sensors on T cells that recognise fragments of pathogens, tumours and the harmless microbes that live in and on us. Each receptor is created by a random genetic shuffling process during T cell development and is then shaped by selection in the thymus, by exposure to antigens and by the environment. The collection of receptors in one person, the TCR repertoire, therefore holds a record of that person’s immune history. Reading this record is difficult: the data are huge, noisy and highly complex. This thesis develops and applies computational frameworks to extract reproducible biological signals from TCR repertoires, with a focus on linking enriched receptor patterns to external factors such as the microbiome and disease. Its central methodological contribution is AIRRWAS (Adaptive Immune Receptor Repertoire-Wide Association Study), a framework that systematically links TCR repertoire patterns to the composition of the gut microbiome. Applied to three independent cohorts, covering healthy individuals, colorectal cancer patients and people with inflammatory bowel disease, AIRRWAS identified reproducible associations between TCRs and specific gut bacteria. Targeted bacterial stimulation experiments confirmed these predictions, making this, to our knowledge, the first framework to predict and functionally validate TCR-bacteria interactions at this resolution. The results show that the gut microbiome leaves a measurable, partly shared imprint on both the circulating and the mucosal T cell repertoire. This extends a principle known for persistent viruses such as cytomegalovirus and Epstein-Barr virus to bacteria of the gut. Beyond the microbiome, the thesis explores TCR signatures in Crohn’s disease, cancer immunotherapy and susceptibility to herpes zoster. In all these settings, TCR convergence, the repeated emergence of similar receptors through independent recombination events, proves to be a powerful principle: it separates antigen-driven signals from background recombination bias, even in the presence of technical noise and differences between cohorts. By combining convergence analysis, TCR clustering, epitope prediction and multimodal data integration, this work shows that TCR repertoires can be read not as lists of individual sequences, but as system-level records of immunological experience. This perspective has broad implications for immune monitoring, biomarker discovery and our understanding of adaptive immunity in health and disease.