Mycobacterium avium subsp. paratuberculosis (MAP), the aetiological agent of Johne's disease, is one of the most important bacterial pathogens in ruminants. A thorough understanding of MAP pathogenesis is needed to develop new vaccines and diagnostic tests. The generation of comprehensive random transposon mutant libraries is a fundamental genetic technology to determine the role of genes in physiology and pathogenesis. In this study, whole MAP genome analysis compared the insertion sites for the mycobacterial transposon Tn5367 derived from the Mycobacterium smegmatis insertion sequence IS1096 and the mariner transposon MycoMarT7 carrying the Himar1 transposase. We determined that only MycoMarT7 provides a random representation of insertions in 99 % of all MAP genes. Analysis of the MAP K-10 genome indicated that 710 of all ORFs do not possess IS1096 recognition sites, while only 37 do not have the recognition site for MycoMarT7. Thus, a significant number of MAP genes remain underrepresented in insertion libraries from IS1096-derived transposons. Analysis of MycoMarT7 and Tn5367 mutants showed that Tn5367 has a predilection to insert within intergenic regions, suggesting that MycoMarT7 is the more adequate for generating a comprehensive library. However, we uncovered the novel finding that both transposons have loci-dependent biases, with Tn5367 being the most skewed. These loci-dependent transposition biases led to an underestimation of the number of independent mutants required to generate a comprehensive mutant library, leading to an overestimation of essential genes. Herein, we also demonstrated a useful platform for gene discovery and analysis by isolating three novel mutants for each transposon.