In silico evaluation of pharmacokinetic properties and molecular docking for the identification of potential anticancer compounds.

INTRODUCTION

In the pharmaceutical field, the rapid and accurate characterization of physicochemical properties is essential for drug development. In this context, in silico methodologies facilitate the early-stage prediction of ADME (Absorption, Distribution, Metabolism, and Excretion) and toxicity (ADME-Tox) parameters, reducing experimental costs and accelerating the screening of viable drug candidates. Computational approaches such as QSAR, SAR, and QSPR enable the assessment of biological activity and pharmacokinetic behavior.

OBJECTIVE

This study aimed to evaluate the ADME(T) profiles of 58 organic compounds using computational tools, establish predictive models for toxicity, and assess inhibitory potential against the TLK2 kinase domain (PDB: 5O0Y)-a protein implicated in breast cancer and intellectual disability.

METHODOLOGY

Chemical structures were optimized using the MMFF94 force field. ADME-Tox descriptors-including Log P, Log S, Caco-2 permeability, CYP450 interactions, hERG inhibition, LD, and DILI-were calculated using SwissADME and PreADMET. Data analysis included Pearson correlation, PCA, hierarchical clustering, and construction of a cosine similarity network. A Random Forest regression model was implemented to predict LD₅₀ values, and molecular docking simulations were conducted using PyRx and Discovery Studio.

RESULTS

Correlation and PCA analyses revealed key trends, including a strong relationship between Log P and Log D, and groupings based on structural similarity. The Random Forest model demonstrated strong predictive performance for LD (r = 0.8410; RMSE = 0.1112), with five-fold cross-validation confirming robustness. Molecular docking identified several compounds with favorable binding affinities to TLK2, notably compound CC-43, which showed the strongest interaction (-8.2 kcal/mol) and a moderate theoretical toxicity profile (LD = 3.186).

CONCLUSIONS

The integration of ADME(T) profiling, machine learning, and molecular docking provides a comprehensive and reproducible computational strategy for drug discovery. The approach enabled the identification of compounds with favorable pharmacokinetic properties and selective inhibitory potential, supporting compound CC-43 as a promising candidate for further exploration in breast cancer therapeutics.