Establishing disease prediction model using database and cluster based on public TCR repertoires

Abstract

T cell immunity plays a crucial role in the human body's immune functions. Investigating the T-cell receptor (TCR), the antigen recognition site of T cells, provides insights into pathogen-specific immune responses and immunotherapy. TCR acquisition through TCR sequencing is actively utilized in various studies, and TCR databases using public data are being constructed. Leveraging TCR sequencing data, we obtained TCR data from over 10,000 samples in more than 100 studies. We established methods to classify and analyze this data, recording information on diseases, tissues, treatments, and acquiring details on TCR gene usage and CDR3 sequences for each dataset. To validate the obtained data, we tested whether there were significant differences in COVID-19 and melanoma compared to healthy datasets. Additionally, we conducted clustering of CDR3 sequences in some samples, confirming the emergence of disease-specific clusters. Advancing these methods and protocols could enable the management and analysis of extensive disease data, facilitating the creation of disease prediction models for new datasets