Turkish_native_goat_genotypes

Paratuberculosis, caused by <i>Mycobacterium avium</i> subsp. <i>paratuberculosis</i> (MAP), remains a notable One Health concern due to its slow clinical progression, environmental persistence, widespread circulation in ruminant systems, and ongoing interest in its possible zoonotic implications. To characterise MAP exposure in Türkiye’s goat populations, we conducted a nationwide survey covering seven breeds across 36 farms in 11 provinces. High-density SNP genotyping combined with mutual-information–based feature preselection was used to retain informative, non-redundant genomic signals while accommodating both linear and nonlinear patterns typical of complex disease traits.Nine heterogeneous machine-learning models—including penalised linear methods, gradient-boosting algorithms, and interaction-sensitive tree ensembles—were applied to examine genotype–phenotype relationships. An ensemble feature-importance analysis, evaluated through 10,000 permutations, identified 31 FDR-significant SNPs representing markers consistently associated with MAP status. Functional annotation indicated involvement of immune-related processes such as cytokine–receptor signalling, antigen presentation, glycan-mediated T-cell regulation, and NF-κB–linked inflammatory pathways. Partial overlap with mixed linear models and genome-wide McNemar tests suggested that both additive and potential nonlinear components contribute to the observed signal.Overall, these findings outline a coherent genomic pattern associated with MAP susceptibility and demonstrate how machine-learning–integrated genomics may complement conventional approaches in resolving complex trait architecture.