Gossypol, a natural polyphenolic compound derived from Gossypium species, has demonstrated broad anticancer activity; however, its clinical application is limited by poor pharmacokinetic properties and toxicity. This study employed an integrated computational pharmacology approach to evaluate gossypol and its derivatives, identify potential cancer-related target proteins, and elucidate their molecular interactions. ADMET profiling, cytotoxicity prediction, target identification, protein expression and prognostic analysis, and molecular docking were systematically performed. Several gossypol derivatives, particularly anhydrogossypol and gossypolone, exhibited improved drug-likeness, reduced predicted toxicity, favorable anticancer activity, and enhanced selectivity toward cancer cells compared with the parent compound. PASS-based target prediction consistently identified DNA-3-methyladenine glycosylase (MPG), a key enzyme in the base excision repair pathway, as a high-confidence molecular target. Clinical relevance analysis revealed that elevated MPG expressions were associated with unfavorable prognosis and were highly expressed across multiple cancer types, including colorectal, breast, and lung cancers. Molecular docking demonstrated strong binding affinities of selected derivatives within the MPG active site, involving conserved and functionally important residues such as TYR-127, TYR-165, CYS-167, and ARG-182. These findings suggest that gossypol derivatives may exert anticancer effects by modulating MPG-mediated DNA repair mechanisms. Overall, this study highlights MPG as a promising therapeutic target and supports further experimental investigation of optimized gossypol derivatives as potential anticancer agents.