Tabriz University of Medical Sciences Pharmaceutical Sciences 1735-403X 32 3 2026 07 31 Development of 3D-QSAR Models to Predict Inhibition Activity of Selective Histone Deacetylase 1-3 Inhibitors 441 450 10.34172/ps.42810 EN Narges Cheshmazar https://orcid.org/0000-0002-9998-8121 Amirhossein Nasri https://orcid.org/0009-0006-9537-4890 Siavoush Dastmalchi https://orcid.org/0000-0001-9427-0770 Journal Article 10.34172/ps.42810 2025 06 09 2025 09 17 Introduction: Histone deacetylase inhibitors (HDACIs) have attracted researchers’ attention as anti-cancer agents. Designing novel HDAC inhibitors is important in drug discovery field of HDAC due to their high potency, less off-target effects, and good pharmacokinetic and pharmacodynamic profiles. 3D quantitative structure-activity relationship (3D‐QSAR) is a computational method used to design novel compounds considering 3D structure of molecules. Methods: In the current study, we have performed two successive QSAR analyses including a classification based for recognizing selective HDAC1-3 inhibitors from non-selective inhibitors and a 3D-QSAR to predict the potency of inhibitors. To this, initially a classification based QSAR was developed to filter selective HDAC1-3 inhibitors from other isoform-selective or pan inhibitors. Also, a receiver operating characteristics (ROC) analysis was performed for evaluation the goodness of classification model performance. Then, three different 3D-QSAR models were developed specifically for the filtered selective HDAC1-3 inhibitors to assess their selectivity and potency. Results: The generated models revealed that some common structural moieties have positive or negative effect towards the potency of the studied compounds against all three HDAC 1-3 isoforms. The results indicated that out of the identified important variables, a variable named DRY-TIP showing the optimum distance between ZBG and cap group. Furthermore, presence of two HBD groups in ZBD to form pseudoring with a zinc ion present in the active site of the enzyme are essential for exerting inhibitory activity. Using partial least square analysis, a 3D-QSAR model with 5 latent variables was generated with q2 values of internal and external validation equal to 0.62, 0.84, 0.83 and 0.72, 0.90, 0.82 for HDAC1,2,3, respectively. Conclusion: The result of the current study can be used to recognize selective HDAC1-3 inhibitors and predict their potencies with the aim of selective HDAC1-3 inhibitor design.