Genomic Resistome Matrix for Meropenem Resistance Prediction in Klebsiella pneumoniae

Binary gene presence/absence matrices used for training and external validation in the manuscript: "A Leakage-Aware Genomic Prediction Pipeline for Meropenem Resistance in Klebsiella pneumoniae Using Transformer-Based Resistome Representation Learning." Bioinformatics Advances, Oxford University Press, 2026. Training dataset (kp_meropenem_training.csv): 1,268 isolates × 195 binary genomic features. Resistant Phenotype column encodes carbapenem susceptibility (Susceptible: 933, Resistant: 335). Safe deduplication was applied to remove conflicting genome signatures prior to model training. External validation dataset (kp_meropenem_external_validation.csv): 305 isolates from seven independent multinational BioProjects. Includes BioProjectAccession and IsolationCountry metadata columns. BioProject accessions: PRJNA288601, PRJNA278886, PRJNA296771, PRJNA307517, PRJNA312972, PRJEB6574, PRJEB22890. All underlying raw genomic data are publicly available from NCBI and EBI under the BioProject accessions listed above. Feature matrices were derived using PATRIC/BV-BRC resistome annotation pipelines. Code repository: https://github.com/SibelKervanci/kp-meropenem-tabtransformer