MACHINE LEARNING IN THE SEARCH FOR NEW ANTI-TUBERCULOSIS AGENTS: FROM CHEMBL DATABASE SCREENING TO THE CLASSIFICATION OF NATURAL COMPOUNDS
DOI:
https://doi.org/10.11603/2312-0967.2026.3.16483Keywords:
machine learning, in silico research, active pharmaceutical ingredient, XGBoost, ChEMBL, minimum inhibitory concentration, Mycobacterium tuberculosis, RDKit, PythonAbstract
Introduction. The study proposes a sequence of interrelated data-processing steps (pipeline) based on machine learning for the identification of potential inhibitors of Mycobacterium tuberculosis.
Materials and Methods. Using data on organic compounds obtained from the ChEMBL database, a training dataset was generated and characterized using RDKit descriptors. The project was implemented using Python in the Google Colab cloud environment. An XGBoost model was employed for machine learning.
Results and Discussion. Evaluation of the trained model using XGBoost demonstrated high classification accuracy. An analysis of key structural descriptors (in particular, the total van der Waals surface area of atoms with a high positive partial charge; unsubstituted aromatic rings or rings with substituents; the lipophilicity of alkyl and aromatic fragments; carbonyl and piperidine fragments; the van der Waals surface area of atoms with a specific electrotopological state, particularly aromatic and hydrophobic molecular fragments) was performed. The model was subsequently tested on a set of natural compounds. Three potential candidates, CHEMBL437538, CHEMBL108730, and CHEMBL108980, were identified as the most promising compounds for further in vitro research.
Conclusions. Machine learning is an effective tool in silico research. The XGBoost model demonstrates a low propensity for false-positive results; specifically, it blocks molecules lacking antitubercular activity.
References
1. World Health Organization. Global Tuberculosis Report 2025 [Internet]. Geneva: WHO; 2025 [cited 2026 May 27]. Available from: https://www.who.int/teams/global-programme-on-tuberculosis-and-lung-health/tb-reports/global-tuberculosis-report-2025.
2. Chen T, Guestrin C. XGBoost: A scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining; 2016 Aug 13–17; San Francisco, CA. New York: ACM; 2016, p. 785–94. DOI: https://doi.org/10.1145/2939672.2939785
3. Google Colab [Internet]. California: Google LLC; c2026 [cited 2026 Jul 26]. Available from: https://colab.research.google.com/drive/1Cgb9D91z9YEhVU8cGhUMtcdrFf5kkkow?usp=sharing.
4. Geirhos R, Jacobsen JH, Michaelis C, Zemel R, Brendel W, Bethge M, et al. Shortcut learning in deep neural networks. Nat Mach Intell. 2020;2(11):665–73. https://doi.org/10.48550/arXiv.2004.07780. DOI: https://doi.org/10.1038/s42256-020-00257-z
5. RDKit: Open-source cheminformatics software [Internet]. RDKit documentation: Chem.MolSurf module. c2026 [cited 2026 Jul 26]. Available from: https://www.rdkit.org/docs/source/rdkit.Chem.MolSurf.html.
6. Labute P. A widely applicable set of molecular descriptors. J Mol Graph Model. 2000;18(4–5):464–77. https://doi.org/10.1016/s1093-3263(00)00068-1. DOI: https://doi.org/10.1016/S1093-3263(00)00068-1
7. Koštrun S, Munic Kos V, Matanović Škugor M. Around the macrolide – Impact of 3D structure of macrocycles on lipophilicity and cellular accumulation. Eur J Med Chem. 2017 Jun 16;133:351–364. https://doi.org/10.1016/j.ejmech.2017.03.056. DOI: https://doi.org/10.1016/j.ejmech.2017.03.056
8. Nikaido H. Molecular basis of bacterial outer membrane permeability revisited. Microbiol Mol Biol Rev. 2003;67(4):593–656. https://doi.org/10.1128/mmbr.67.4.593-656.2003. DOI: https://doi.org/10.1128/MMBR.67.4.593-656.2003
9. Wildman SA, Crippen GM. Prediction of physicochemical parameters by atomic contributions. J Chem Inf Comput Sci. 1999;39(5):868–73. https://doi.org/10.1021/ci990307l. DOI: https://doi.org/10.1021/ci990307l
10. Hall LH, Mohney BK, Kier LB. The electrotopological state: an atom index for structure-property correlations. J Chem Inf Comput Sci. 1991;31(1):76–81. https://doi.org/10.1021/ci00001a012. DOI: https://doi.org/10.1021/ci00001a012
11. Brennan PJ, Nikaido H. The envelope of mycobacteria. Annu Rev Biochem. 1995;64:2–63. https://doi.org/10.1146/annurev.bi.64.070195.000333. DOI: https://doi.org/10.1146/annurev.bi.64.070195.000333
12. Dessen A, Quemard A, Blanchard JS, Jacobs WR, Sacchettini JC. Crystal structure and function of the isoniazid target of Mycobacterium tuberculosis. Science. 1995;267(5204):1638–43. https://doi.org/10.1126/science.7886450. DOI: https://doi.org/10.1126/science.7886450
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
The authors who publish their materials in this journal agree with the following terms:
- Authors reserve the right to authorship of their work and assign to the journal the right to first publish this work under the terms of the Creative Commons Attribution License, which allows other persons to freely distribute the published work with a mandatory reference to the authors of original work and the first publication of work in this journal
. - Authors have the right to make independent extra-exclusive work agreements in the form they are published by this journal (for example, posting work in an electronic repository of an institution or publishing as part of a monograph), provided that the link to the first publication of the work in this journal is maintained.
Journal policy allows and encourages publication of manuscripts on the Internet (for example, in institutions repositories or on personal websites), both before the publication of this manuscript and during its editorial review, as it contributes to productive scientific discussion and positively affects the efficiency and dynamics of the citation of the published work (see The Effect of Open Access).
