The study introduces LLMTB, a large language model that predicts antibiotic resistance in Mycobacterium tuberculosis by analyzing genomic data. It identifies key resistance genes and intergenic regions associated with various antibiotics, demonstrating superior performance compared to existing methods.