Computational evaluation of chemotherapeutic drugs using topological indices and multicriteria decision analysis

Abstract

This article presents a comprehensive computational framework for analyzing chemotherapeutic drugs through the integration of topological indices and advanced multi-criteria decision-making (MCDM) techniques. Beginning with the construction of molecular graphs for chemotherapy drugs, various topological indices are computed using Python programming to characterize and quantify the molecular structure. We applied Quantitative Structure–Property Relationship (QSPR) modeling to predict important physicochemical properties, such as density, molar refractivity, polarizability, and flash point, on the basis of these indices, demonstrating the efficacy of the indices. Although each constituent tool—degree- and distance-based indices, Shannon–entropy weighting, COPRAS, TOPSIS, and polynomial regression—is classical and extensively validated, our contribution is to specify a single, auditable pipeline that records graph construction choices, normalization, weighting, and ranking steps in one place, thereby reducing ambiguities that often arise when descriptors and decision rules are reported separately. The study further incorporates MCDM methods, specifically the Complex Proportional Assessment (COPRAS) and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), to rank and prioritize these drugs by integrating multiple topological and physicochemical criteria. Detailed statistical analysis, including entropy weighting and correlation studies, validates the robustness of the models and rankings. The correlation coefficient between the topological indices and the physicochemical characteristics of drugs is calculated using SPSS software. Shannon’s entropy method is used to determine the weight of different criteria. Furthermore, three-dimensional visualization techniques illustrate the comparative rankings obtained from COPRAS and TOPSIS methods. Those integer ranks are nevertheless authoritative only in tabular form (Tables 13 and 19); the 3D graphic is an optional qualitative juxtaposition and does not define or refine the underlying orderings. We articulate the following principal contributions: (i) a unified chemotherapy-focused case study that couples the selected topological descriptors with entropy-weighted COPRAS and TOPSIS scores under identical input matrices; (ii) a side-by-side discussion of agreement and discordance between these rankings, together with correlation- and regression-based diagnostics that identify which indices track key physicochemical endpoints in our panel; and (iii) explicit documentation intended to support reproducibility of the numerical workflow beyond ad hoc calculator use. The results demonstrate that graph-theoretic descriptors combined with entropy-weighted MCDM provide a transparent computational framework for the structural prioritization of chemotherapeutic compounds according to the selected graph-based and physicochemical criteria. The proposed workflow is intended as an early-stage chemoinformatics decision-support tool and should not be interpreted as a substitute for biological, pharmacological, toxicological, or clinical evaluation. Descriptor-driven multi-criteria ranking for drug-like compounds is now well documented in the cheminformatics literature. In that sense, “potential” refers to transparent computational prioritization under the stated structural and physicochemical criteria, not to unsupervised clinical prescribing. With respect to ADMET (absorption, distribution, metabolism, excretion, toxicity), contemporary profiling explicitly multiplexes solubility, permeability, metabolic stability, clearance, protein binding, and organ-toxicity surrogates; those endpoints are not modeled by the present graph-only criterion set, and their inclusion is left as mandatory future work before any regulatory or bedside interpretation could be entertained. We therefore position the manuscript as computational decision support grounded in transparent descriptors, without claiming new axioms for MCDM or new index formulas beyond the staged integration emphasized above. This methodological synthesis provides a significant contribution to cheminformatics and computational drug discovery.