The species comprises six subtypes that were recently classified into six closely related species; (formerly subtype 1), (subtype 2), (subtype 3), (subtype 4), (subtype 5) and (subtype 6). Together with , they form the complex. is the most frequent and most pathogenic species of the complex. is classically associated with diseases in immunosuppressed patients, and the other species are mostly colonizers, and are only very rarely reported in ill patients. Comparative genomics was used to assess the genetic determinants leading to the pathogenicity of members of the complex. The genomes of 51 isolates collected from patients with and without disease were sequenced and compared with 24 publicly available genomes. The pathogenicity of each isolate was determined based on the clinical records or public metadata. A comparative genomic analysis showed that all , , and isolates lacked the ESX-1-associated EspACD locus that is thought to play a crucial role in the pathogenicity of and other non-tuberculous mycobacteria. Furthermore, was the only species exhibiting a 25-Kb-large genomic island encoding for 17 type-VII secretion system-associated proteins. Finally, a genome-wide association analysis revealed that two consecutive genes encoding a hemerythrin-like protein and a nitroreductase-like protein were significantly associated with pathogenicity. These two genes may be involved in the resistance to reactive oxygen and nitrogen species, a required mechanism for the intracellular survival of bacteria. Three non-pathogenic lacked these genes likely due to two distinct distributive conjugal transfers (DCTs) between and , and one DCT between and . To our knowledge, this is the first study linking DCT to reduced pathogenicity.