Whole-genome sequencing is assuming an increasingly important role in guiding the clinical care of patients with tuberculosis. While bioinformatics resources for processing and clinically interpreting Mycobacterium tuberculosis genomic data are advancing, a persistent gap remains for user-friendly, adaptable tools that can be readily customized for local clinical reporting requirements and meet the rigorous demands of international laboratory accreditation. This study aimed to develop a dependable software application to determine M. tuberculosis lineages and antimicrobial resistance patterns directly from genomic sequences, engineered for clinical use and accessible to routine microbiology laboratories. We created tbtAMR, a versatile and comprehensive data-driven platform for analyzing M. tuberculosis genomic data, encompassing both lineage identification and phenotypic drug-susceptibility prediction. The tool processes short-read sequencing data in FASTQ format or annotated VCF files derived from short- or long-read technologies. It systematically detects genomic variants—including single-nucleotide polymorphisms, insertions, deletions, large structural rearrangements, and gene loss or inactivation—and cross-references resistance-conferring mutations against the WHO catalog (or a custom database). It generates a structured report that classifies resistance profiles for clinical use. We validated it through extensive benchmarking against phenotypic and genomic results from our in-house collection (n = 2005) and large external repositories and published datasets (n = 13 777). We also tested simulated sequence data with deliberately introduced known variants to establish optimal quality thresholds and confirm the pipeline’s robustness for diagnostic deployment. We compared performance against established public tools, notably TBProfiler and Mykrobe.tbtAMR delivered precise lineage assignments and reliable predictions of phenotypic susceptibility for first-line agents (sensitivity 94.6% [95% CI: 94.2–95.0], specificity 97.5% [97.3–97.7]) and second-line agents (sensitivity 83.7% [82.7–84.7], specificity 98.0% [97.9–98.1]). Its analytical speed and overall accuracy matched or approached those of other widely used bioinformatics platforms, such as TBProfiler (first-line sensitivity 94.2% [93.0–95.3], specificity 97.9% [97.6–98.2]) and Mykrobe (first-line sensitivity 91.5% [90.0–92.8], specificity 98.4% [98.2–98.6]). The software further provides configurable parameters that enable laboratories to adjust reporting thresholds according to their individual clinical contexts.tbtAMR represents a practical solution for clinical and public health microbiology laboratories, offering straightforward customization by laboratory staff without specialized programming skills. The tool has successfully achieved ISO accreditation within our setting. It is currently integrated into routine workflows for genomic antimicrobial resistance reporting, delivering results in a timeframe comparable to conventional phenotypic testing (typically 3–4 weeks from positive culture). Comprehensive supporting materials—including reporting templates, detailed validation frameworks, and reference datasets—are provided to help other laboratories implement and accredit this important diagnostic approach, ultimately supporting enhanced tuberculosis control worldwide.
