Antibiotics resistance is a clear threat to the future of current tuberculosis treatments like rifampicin, prompting the need for new treatment options in this field. While plants can offer a plethora of chemical diversity in their constitutive natural products to tackle this issue, finding potentially bioactive compounds in them has not always proven to be that simple. Classical bioactivity-guided fractionation approaches are still trendy, but they bear significant shortfalls, like their time-consuming nature as well as the ever-increasing risk of isolating known bioactive compounds. In this regard, we have developed an alternative method to the latter approach that allows for natural derivatives of a known bioactive scaffold to be efficiently targeted and isolated within a large library of plant extracts. Hence our approach allows for the anticipation of bioactive structure independently of preliminary bioassays. By relying on the chemical diversity of a set of 1,600 plant extracts analyzed by HRMS/MS, we were able to isolate and characterize several minor derivatives of a previously reported bioactive aza-anthraquinone compound from Cananga brandisiana , selected within the plant set. Assessment of bioactivity on these derivatives (especially onychine, with an IC ₅₀ value of 39 µM in infection) confirmed their expected activity on Mycobacterium marinum in our anti-infective assay. This proof-of-concept study has established an original path towards bioactive compounds isolation, with the advantage of potentially highlighting minor bioactive compounds, whose activity may not even be detectable at the extract level.