Résumé:
This study was conducted to support the ongoing search for novel molecules effective
enough to treat breast cancer derived from Trigonella foenum-graecum, Curcuma longa L.,
and Atriplex Halimus. The study combined computational techniques and computer
simulations to evaluate their efficacy, as well as Fourier transform infrared (FTIR) and
ultraviolet/visible (UV/VIS) spectroscopy of plant extract. Quantitative structure-activity
relationship (QSAR) modeling was used to predict bioactivity; with 2D_3D-QSAR models
developed using multiple linear regression (MLR) and support vector regression (SVR).
Comparative analysis revealed that MLR outperformed SVR, achieving an R² value of 0.93,
an RMSE value of 0.163, and a high Q² value of 0.87, demonstrating superior predictive
accuracy. Model predictions were validated using cross-validation. Molecular docking
simulations evaluated the binding interactions with key breast cancer targets (HER2, CDK4,
AKT1, and MCF-7), revealing strong affinities ranging from -7 to -11.4 kcal/mol.
Furthermore, in conjunction with the density functional theory (DFT) method (B3LYP/6-
311G(d,p)), molecular docking identified two promising candidates, fenugreekine and
diosgenin, derived from fenugreek, which demonstrated potent inhibition of all proteins
compared to the FDA-approved drugs Ibrance and Capivasertib, respectively. ADMET
(absorption, distribution, metabolism, excretion, and toxicity) analysis of diosgenin,
curcumin, and arbutin demonstrated their pharmacokinetic affinity and safety, making them
promising candidates for further experimental validation
