Chromatic hyper Zagreb topological descriptors and QSPR analysis of certain skin cancer drugs Authors: R. Pavithra, J. Ravi Sankar Published: 2026 ABSTRACT Skin cancer is one of the most common diseases, and studying its drugs is important for effective treatment and drug design. Drugs for melanoma, immunosuppressants, and topical drug administration have a bright future in the investigation of skin cancer drugs with possible anticancer properties. The use of chromatic topological descriptors is still the predominant strategy due to the notable progress being made in the field of drug creation. In this work, the investigation of regression models and chromatic hyper Zagreb topological descriptors for some skin cancer drugs is the main focus of this study. Together with the QSPR models, descriptors provide a numerical representation of a molecule’s chemical characteristics. Proper coloring is applied to the molecular graphs such that no two adjacent vertices share the same color. Regression models are constructed for the calculated index values while the physicochemical properties of skin cancer drugs are investigated. The framework that connects chemical composition to physical attributes is represented by numbers associated with chromatic topological indices. Based on the data collected, an analysis is conducted for a number of noteworthy findings. Introduction The biggest organ in the human body is the skin, which acts as a barrier to protect the body from external dangers, including infections, dangerous UV rays, and chemical exposure. There are three main layers, which are the epidermis, the outermost layer, which is made up of keratinocytes that give structure and melanocytes that produce pigment (Maranke, 2009). The second is the dermis, an intermediate layer that contains sweat glands, connective tissues, blood vessels, nerve endings, and hair follicles, and the last is the hypodermis (subcutaneous tissue), the body’s lowest layer, made up of connective tissues and fat, which aids in insulating and cushioning. In addition to avoiding dehydration and facilitating sensory awareness, the skin is essential for controlling body temperature. However, cellular damage can result from extended exposure to dangerous UV radiation from the sun or artificial sources and increase the risk of skin cancer (Queen, 2017). Skin cancer is characterized by aberrant skin cell development that results from DNA alterations that cause unchecked cell division. It is one of the most prevalent forms of cancer on the globe and comes in three main varieties: BCC is the most prevalent and least dangerous kind of skin cancer that arises in the epidermis’ basal cells (Dika et al., 2020). The squamous cells in the epidermis give birth to squamous cell carcinoma (SCC), which is more likely to spread than BCC. Originating from melanocytes, melanoma is the most aggressive and lethal kind of skin cancer (Hunter Shain and Bastian, 2016). In the event that melanoma is not identified and treated promptly, it is likely to spread. Moreover, cutaneous lymphoma and merkel cell carcinoma are less frequent types (Becker et al., 2017). Knowing a drug’s physicochemical, biological, and pharmacological characteristics requires a comprehension of its molecular structure. In graph theory, a graph G=(V,E) is an ordered pair consisting of a non-empty set of vertices V (or nodes/points) and a set of edges E (lines or arcs) that connect pairs of vertices (Zhang and Chartrand, 2006). It represents pairwise relationships between objects, where edges define connections between vertices. In graph theory, a simple graph is an undirected graph that contains no loops and no multiple edges. It is the most basic representation of relationships, where each pair of vertices is connected by at most one edge. Topological indices are numerical invariants derived from molecular graphs, which represent atoms as vertices and bonds as edges that quantify structural features to predict physicochemical properties (Balaban and Ivanciuc, 2000). They offer important information on molecular similarities, activity prediction, and drug design. In chemical graph theory, QSAR is a method that relates the biological activity and molecular structure of a compound whereas QSPR relates a mathematical relationship between the molecular structure of a compound and its physicochemical properties. In chemistry, degree-based topological indices are numerical graph invariants that are used to connect molecule structures with physical and chemical properties such as stability and boiling point without the need for synthesis. Common indices used in QSAR/QSPR investigations are the Randić index, Zagreb indices, and atom-bond connectivity indices, which are calculated by summing functions of vertex degrees (dv) (Gutman, 2013). Chemical graph theory is a subfield of mathematical chemistry that applies graph theory to represent and analyze molecular structures, where vertices denote atoms and edges represent chemical bonds (Trinajstic, 2018). Chemical graph theory, drug design, and computational chemistry all make extensive use of topological indices, which offer a mathematical method for examining molecular structures. Colakoglu talks about past research on potential drugs to treat COVID-19 (Çolakoğlu, 2022). Given that it is a costly and intricate phenomenon, this method is most effective when used to forecast discoveries. By using QSPR modeling, Nasir et al.’s blood cancer medication results in demonstrate a substantial relationship between pharmacological properties and TDs (Nasir et al., 2022). Drug design, molecular similarity analysis, and QSAR/QSPR research all make extensive use of distance-based topological indices. Here are some results with topological descriptors in molecular graphs. Two regression methods were examined by Pandeeswari et al. to determine the precision of their forecast by analyzing topological indicators in medications for breast cancer (Pandeeswari and Ravi Sankar, 2025; Pandeeswari and Sankar J, 2025). They offer crucial information on reaction processes, toxicity prediction, boiling temperatures, and molecule stability. Graph energy and topological descriptors of zero divisor graphs was determined by Clement Johnson et al., which associate with commutative rings and provide a strong bridge between spectral graph theory and algebraic structure (Johnson Rayer and Jeyaraj, 2023; Johnson and Sankar, 2023; Clement Johnson Rayer and Ravi Sankar Jeyaraj, 2023). Simran Kour et al. combine machine learning regression approaches and examine a characterization of tricyclic antidepressant drugs and anti-cancer drugs properties by capturing complex relationships between biological activity and molecular descriptors (Kour and Sankar, 2025; Kour and Sankar, 2024). C. Glory et al. investigate b-chromatic and rainbow chromatic topological indices for various graphs and their derived and central structures, capturing significant relationships between graph coloring techniques and structural properties (Glory and Nanjappa, 2023; Glory et al., 2024). C. Yogalakshmi et al. develop a QSPR graph model using novel coloring-based topological indices to explore physicochemical properties of potential antiviral drugs, capturing complex relationships between molecular structure and drug activity (Yogalakshmi and Balamurugan, 2025). Based on the lower bound for the second hyper-Zagreb index of trees with a given Roman domination number, providing key theoretical insights into graph invariants and structural properties established by Ali et al. (2025)). R. Rajambigai et al. examine the structure–efficiency relationship of access group antibiotics via SK chromatic descriptors, effectively linking molecular topology with antibiotic performance (Rajambigai et al., 2026). In particular, we highlight the fundamental molecule structure that has a greater direct influence on molecular characteristics and biological processes by using hydrogen-depleted graphs, which do not include hydrogen atoms (Gross et al., 2018). This study aims to develop regression models for predicting QSPR based on physicochemical properties while simultaneously computing chromatic hyper Zagreb topological indices for 10 skin cancer drugs using proper coloring. The reliability and predictive accuracy of these models are assessed by systematically comparing calculated and experimental values. Classical topological indices are widely used to describe molecular structures based on connectivity patterns. However, chromatic Zagreb indices extend these descriptors by including vertex coloring, which provides additional structural information about the molecule. By integrating chemical graphs with molecular descriptor analysis, this work provides a comprehensive topological and physicochemical characterization of the compounds. Moreover, the study exhibits the adaptability of this approach across different classes of skin cancer drugs, highlighting its potential as a powerful tool for drug design and structure–activity relationship studies, and it encourages the creation of new drugs for potential treatment. Formulas and methods Ten skin cancer drugs were selected for analysis in this study, and the PubChem (National Center for Biotechnology Information NCBI, 2026) and ChemSpider databases (ChemSpider, 2025) were used to gather information on their physicochemical properties. The drugs considered in this study, namely, Tirbanibulin, Alitretinoin, Methotrexate, Tazarotene, Nicotinamide, Diclofenac, Imiquimod, Topotecan, Vorinostat, and Sonidegib, are commonly used in the treatment of skin cancer and related dermatological conditions. Table 1 shows the experimental data for the physical and chemical characteristics of the ten skin cancer drugs, which are shown in Figure 1. As a result, the regression model is a good test to use and examine the data. Figure 2 illustrates the proper coloring of the molecular graphs corresponding to certain skin cancer drugs. This step is essential for the computation of chromatic topological indices, as these indices depend on the assignment of colors to vertices such that no two adjacent vertices share the same color. TABLE 1Skin cancer drugs with their physicochemical properties.DrugMolecular weight (MW)Molar refractivity (MR)Polarizability (P)Complexity (C)Polar surface area (PSA)Molar volume (MV)Surface tension (ST)log PTirbanibulin431.5124.249.254064369.048.71.97Alitretinoin300.495.537.956737297.239.16.83Methotrexate454.5119.047.2704211295.796.5−0.24Tazarotene351.5101.440.254764287.057.56.22Nicotinamide122.233.513.31145799.150.2−0.24Diclofenac318.176.530.331049206.858.04.06Imiquimod240.371.028.229457187.748.83.46Topotecan421.4112.744.7867103281.382.91.08Vorinostat264.373.529.127678225.050.40.86Sonidegib485.5126.750.269164386.843.15.42 Skin cancer drugs with their physicochemical properties. Molecular structures of skin cancer drugs. (a) Tirbanibulin. (b) Alitretinoin. (c) Methotrexate. (d) Tazarotene. (e) Nicotinamide. (f) Diclofenac. (g) Imiquimod. (h) Topotecan. (i) Vorinostat. (j) Sonidegib. Proper coloring of molecular graphs of skin cancer drugs. (a) Tirbanibulin. (b) Alitretinoin. (c) Methotrexate. (d) Tazarotene. (e) Nicotinamide. (f) Diclofenac. (g) Imiquimod. (h) Topotecan. (i) Vorinostat. (j) Sonidegib. Chromatic modified and chromatic hyper Zagreb indices The formula was provided in our earlier unpublished research paper (Pavithra, 2026). Chromatic modified first Zagreb index: M1ΦmG=∑v∈VG1cv2=∑j=1l1θcj j2 (1) Chromatic modified second Zagreb index: M2ΦmG=∑uv∈EG1cucv=∑1≤t,s≤χGt 0.7). Correlation coefficient of physicochemical properties. Bold value indicate strong correlation coefficients (r > 0.7). Figure 3 displays the graph of the correlation coefficient of chromatic modified and chromatic hyper Zagreb indices and all the physicochemical properties, including molecular weight, molar refractivity, polarizability, complexity, polar surface area, molar volume, surface tension and log P. To evaluate the relationship between physicochemical properties and chromatic topological indices, correlation coefficients were created. Graphical representation of the correlation coefficient and physicochemical properties of the drugs. Standard estimate error and statistical metrics evaluation The statistical parameters are combined for all physicochemical properties and chromatic topological indices in Tables (5–8) to assist us in understanding how they relate to each other. A number of statistical characteristics are observed, including the significant value p, Fisher’s statistic, the percentage of dependent variables R2 , the slope b, the constant term A, the sample size N, and the importance of the association for each of the chromatic topological indices. From an interpretive standpoint, a p-value less than 0.05 is considered statistically significant, whereas a p-value of more than 0.05 is seen as not being statistically significant. TABLE 5Statistical parameters of the linear QSPR model using M1Φm(G) .PropertiesNAB R2 FpIndicatorMW10497.35−1145.540.5439.5120.015SignificantMR10135.87−307.190.564610.370.012SignificantP1053.83−121.530.56310.310.012SignificantC10718.40−1644.750.2572.770.134InsignificantPS1099.99−156.200.0510.4370.526InsignificantMV10397.03−965.370.66115.600.004SignificantST1059.37−13.380.00280.0220.883Insignificantlog P104.794−13.390.1361.2600.294Insignificant Statistical parameters of the linear QSPR model using M1Φm(G) . TABLE 6Statistical parameters of the linear QSPR model using M2Φm(G) .PropertiesNAB R2 FpIndicatorMW10351.21−114.900.0330.2790.611InsignificantMR1096.29−27.110.02710.2230.6493InsignificantP1038.164−10.630.0260.2180.652InsignificantC10489.1117.700.00010.00140.97InsignificantPS1079.98−14.860.00280.02320.882InsignificantMV10278.37−138.990.0840.7380.415InsignificantST1056.2411.940.0140.1140.7443Insignificantlog P103.109−1.570.01150.0930.767Insignificant Statistical parameters of the linear QSPR model using M2Φm(G) . TABLE 7Statistical parameters of the linear QSPR model using HM1Φ(G) .PropertiesNAB R2 FpIndicatorMW1044.791.2290.91687.540.00001SignificantMR1016.470.32140.905776.860.00002SignificantP106.540.1270.90677.2030.00002SignificantC10−92.732.4390.82939.040.0002SignificantPS1017.550.2540.2012.0160.193InsignificantMV1062.240.8410.73522.220.0015SignificantST1036.230.08890.1851.8170.214Insignificantlog P102.4050.0020.00550.0440.837Insignificant Statistical parameters of the linear QSPR model using HM1Φ(G) . TABLE 8Statistical parameters of the linear QSPR model using HM2Φ(G) .PropertiesNAB R2 FpIndicatorMW1087.772.22440.73822.550.001SignificantMR1027.530.5880.73322.030.001SignificantP1010.9180.2330.73422.140.001SignificantC10−47.844.8150.78128.590.0006SignificantPS1025.1770.4750.1701.6400.236InsignificantMV10100.541.4560.5329.120.016SignificantST1035.910.1930.2102.1350.182Insignificantlog P102.7440.00170.00080.0060.936Insignificant Statistical parameters of the linear QSPR model using HM2Φ(G) . These measures not only enable comparisons but also provide accurate computations. The physicochemical properties of the skin cancer drugs are listed with the standard estimation error in Table 9. The accuracy of the QSPR model predictions are facilitated by the computation of standard estimation error. The comparison between the calculated and observed values of physicochemical properties, based on the regression models of chromatic modified and chromatic hyper Zagreb indices, are presented in Tables 10–17. The correlation determination provides more information about the relationship between variables by describing the percentage of the association. TABLE 9Standard estimate error for physicochemical properties of the drugs.DrugMolecular Weight (MW)Molar Refractivity (MR)Polarizability (P)Complexity (C)Polar surface Area (PSA)Molar Volume (MV)Surface Tension (ST)log P M1Φm(G) 80.8720.768.242214.9751.4153.2119.232.59 M2Φm(G) 117.6231.0412.30249.4752.7287.4519.122.779 HM1Φ(G) 34.629.663.82102.8847.1847.0217.392.787 HM2Φ(G) 61.2316.246.423116.6548.0962.4817.112.794 Standard estimate error for physicochemical properties of the drugs. TABLE 10Comparison of observed and predicted values of molecular weight based on regression models.DrugMolecular weight M1Φm(G) M2Φm(G) HM1Φ(G) HM2Φ(G) Tirbanibulin431.5407.855349.569431.925401.930Alitretinoin300.4379.712348.602288.132285.242Methotrexate454.5408.915349.569431.925401.930Tazarotene351.5389.546349.142343.437330.122Nicotinamide122.2196.395344.815144.339168.554Diclofenac318.1357.393348.338266.010267.290Imiquimod240.3206.997290.571294.277350.318Topotecan421.4274.609311.125468.795495.778Vorinostat264.3351.939348.189254.949258.314Sonidegib485.5416.875349.700465.208428.658 Comparison of observed and predicted values of molecular weight based on regression models. TABLE 11Comparison of observed and predicted values of molar refractivity based on regression models.DrugMolar refractivity M1Φm(G) M2Φm(G) HM1Φ(G) HM2Φ(G) Tirbanibulin124.2111.87195.903117.711109.850Alitretinoin95.5104.32895.67580.10779.274Methotrexate119.0112.15595.903117.711109.850Tazarotene101.4106.96595.80294.58091.034Nicotinamide33.555.22794.78342.50348.698Diclofenac76.598.33995.61274.32274.570Imiquimod71.057.99581.98481.71496.326Topotecan112.776.14886.831127.353134.546Vorinostat73.596.62395.57771.42972.218Sonidegib126.7114.28995.934126.389116.906 Comparison of observed and predicted values of molar refractivity based on regression models. TABLE 12Comparison of observed and predicted values of polarizability based on regression models.DrugPolarizability M1Φm(G) M2Φm(G) HM1Φ(G) HM2Φ(G) Tirbanibulin49.244.33538.01246.54543.538Alitretinoin37.941.35137.92331.68631.422Methotrexate47.244.44838.01246.54543.538Tazarotene40.242.39637.97337.40136.082Nicotinamide13.321.93037.57316.82719.306Diclofenac30.338.98337.89829.40029.558Imiquimod28.223.02532.55332.32138.179Topotecan44.730.20634.45550.35553.324Vorinostat29.138.30137.88428.25728.626Sonidegib50.245.29138.02449.97446.334 Comparison of observed and predicted values of polarizability based on regression models. TABLE 13Comparison of observed and predicted values of complexity bases on regression models.DrugComplexity M1Φm(G) M2Φm(G) HM1Φ(G) HM2Φ(G) Tirbanibulin540589.888489.363675.555626.260Alitretinoin567549.522489.512390.192375.880Methotrexate704591.408489.363675.555626.260Tazarotene547563.616489.429499.947472.180Nicotinamide114286.741490.094104.829125.500Diclofenac310517.467489.553346.290337.360Imiquimod294301.578498.452402.387515.515Topotecan867398.653495.285748.725828.490Vorinostat276508.214489.575324.339317.100Sonidegib691602.822489.343741.308684.040 Comparison of observed and predicted values of complexity bases on regression models. TABLE 14Comparison of observed and predicted values of polar surface area based on regression models.DrugPolar surface area M1Φm(G) M2Φm(G) HM1Φ(G) HM2Φ(G) Tirbanibulin6487.78779.76897.56091.677Alitretinoin3783.95079.64367.84266.977Methotrexate21187.93079.76897.56091.677Tazarotene6485.29279.71379.27276.477Nicotinamide5758.97879.15438.12442.277Diclofenac4980.90879.60963.27063.177Imiquimod5760.39372.13569.11280.752Topotecan10369.62874.795105.180111.627Vorinostat7880.03179.58961.98461.277Sonidegib6489.01479.785104.41897.377 Comparison of observed and predicted values of polar surface area based on regression models. TABLE 15Comparison of observed and predicted values of molar volume based on regression models.DrugMolar volume M1Φm(G) M2Φm(G) HM1Φ(G) HM2Φ(G) Tirbanibulin369.0321.611276.385327.155304.380Alitretinoin297.2297.912275.215228.758228.668Methotrexate295.7322.502276.385327.155304.380Tazarotene287.0306.176275.868266.603257.788Nicotinamide99.1143.606270.642130.361152.956Diclofenac206.8279.101274.895213.620217.020Imiquimod187.7152.379205.016232.963270.892Topotecan281.3209.379229.879352.385365.532Vorinostat225.0273.674274.714206.051211.196Sonidegib386.8329.192276.542349.962321.852 Comparison of observed and predicted values of molar volume based on regression models. TABLE 16Comparison of observed and predicted values of surface tension based on regression models.DrugSurface tension M1Φm(G) M2Φm(G) HM1Φ(G) HM2Φ(G) Tirbanibulin48.758.32556.41064.23462.93Alitretinoin39.157.99656.51153.83252.89Methotrexate96.558.33756.41064.23462.93Tazarotene57.558.11156.45557.83356.75Nicotinamide50.255.85856.90443.43142.85Diclofenac58.057.73656.53852.23251.35Imiquimod48.855.97862.54354.27758.49Topotecan82.956.76960.40666.90171.03Vorinostat50.457.66056.55451.43250.57Sonidegib43.158.43056.39766.63465.24 Comparison of observed and predicted values of surface tension based on regression models. TABLE 17Comparison of observed and predicted values of log⁡P based on regression models.Druglog P M1Φm(G) M2Φm(G) HM1Φ(G) HM2Φ(G) Tirbanibulin1.973.7483.0873.0352.982Alitretinoin6.833.4193.0732.8012.894Methotrexate−0.243.7603.0873.0352.982Tazarotene6.223.5343.0812.8912.928Nicotinamide−0.241.2793.0222.5672.805Diclofenac4.063.1583.0702.7652.880Imiquimod3.461.4002.2802.8112.943Topotecan1.082.1922.5613.0953.053Vorinostat0.863.0833.0682.7472.873Sonidegib5.423.8543.0883.0893.002 Comparison of observed and predicted values of log⁡P based on regression models. Discussion In this work, we studied the relationship between 8 physicochemical characteristics of ten skin cancer drugs with the four chromatic Zagreb topological indices. Table 4 represents the correlation coefficients between the four chromatic Zagreb topological indices and the eight physicochemical properties, with particular attention paid to the substantial association (r>0.7) that was observed. After looking at everything, the properties showed the moderate to strong correlation with the chromatic modified and chromatic hyper Zagreb indices, with molecular weight (r = 0.957), molar refractivity (r = 0.951), polarizability (r = 0.951), complexity (r = 0.911), molar volume (r = 0.857), which indicates that these chemical characteristics may be well predicted by these chromatic topological indices. These physicochemical properties show negligible because all of the chromatic topological indices exhibit a weak association (r<0.2) with the polar surface area, surface tension and log P. Tables 5–8 presents the QSPR analysis utilizing statistical features such R2 , F-statistic, p-value, and regression coefficients A and B. Notably, molecular weight has R2=0.916 , F=87.54 , p=0.00001 , polarizability has R2=0.906 , F=77.203 , p=0.00002 , molar refractivity has R2=0.9057 , F=76.86 , p=0.00002 , with HM1Φ(G) index. Remaining some physical properties shows the moderate dependency on chromatic topological indices such as complexity has R2=0.829 , F=39.04 , P=0.0002 and molar volume has R2=0.735 , F=22.22 , p=0.0015 with HM1Φ(G) index. The polar surface area, surface tension and log P show weaker correlations compared to other properties. This may be due to the fact that these properties depend on complex molecular interactions and environmental factors that are not fully captured by the selected chromatic topological descriptors. Table 6 shows the QSPR model based on the M2Φm(G) index shows insignificant relationships with all the considered physicochemical properties, indicating that this index has limited applicability for predictive modeling in the present study. While examining the correlation determination for physical properties under consideration, we have that HM1Φ(G) index gives the moderate to strong correlation determination for molecular weight (R2=0.916) , molar refractivity (R2=0.9057) and polarizability (R2=0.906) , then complexity (R2=0.829) , molar volume (R2=0.735) . We can also see high percentage correlation between chromatic modified and chromatic hyper Zagreb index and physical properties of drugs. HM1Φ(G) index has positive and high percentage of correlation for molecular weight (R2=0.916) , molar refractivity (R2=0.9057) and polarizability (R2=0.906) . According to this analysis, the values of the chromatic modified and chromatic hyper Zagreb indices are essential for comprehending the properties that are predicted, such as molecular weight, that helps in comprehending the intermolecular forces that aid in the design of models with particular dielectric properties. Drug design can also benefit from the use of the chromatic modified and chromatic hyper Zagreb index values for estimating molecular weight, molar refractivity, and polarizability, for which the predicted error is quite large. Limitations and future directions This research demonstrates significant correlations between chromatic topological indices and selected physicochemical properties; certain limitations should be acknowledged. First, the analysis is based on a relatively small dataset of ten skin cancer drugs, which may limit the generalizability of the results. Second, the study primarily employs linear regression models, which may not fully capture complex non-linear relationships between molecular descriptors and physicochemical properties. Furthermore, some properties such as non-linear models or hybrid modeling may need to be investigated for greater accuracy to predict polar surface area, surface tension, and log P. The sample size considered in this study is relatively small, which may limit the general applicability of the developed QSPR models. A larger dataset could provide more reliable and generalized results. Future research may focus on incorporating additional statistical parameters such as Root Mean Square Error (RMSE) and Average Absolute Relative Deviation (AARD) to provide a more detailed evaluation of the QSPR models. Further studies can also explore advanced regression techniques such as quadratic and cubic models and incorporate machine learning approaches to improve predictive accuracy. In order to enhance prediction and guide future research, the study therefore demonstrates the necessity of adjusting QSPR models to characterize physicochemical properties. These directions can significantly contribute to the development of more accurate and efficient computational models for drug design and analysis. Conclusion To comprehend the structural characteristics of skin cancer drugs, four chromatic Zagreb topological indices have been used. The physicochemical properties of some drugs have also been subjected to QSPR investigation. Certain chromatic topological indices are shown to be effective in predicting characteristics like polarizability, molecular weight, and molar refractivity. Thus, the work demonstrates that molecular structure is crucial in controlling the properties of specific drugs, suggesting that chromatic Zagreb topological indices are required to predict these characteristics. The estimated value produced from this will enable the pharmaceutical industry to develop new drugs that will surely be helpful in obtaining preventative measures for the aforementioned illness. For new exposures to different illnesses, they offer methods for estimating characteristics. Entropy, critical pressure, boiling point, acentric factor, enthalpy, and other characteristics and processes may all be determined and predicted with its help. Additionally, our findings may help in the creation of new drugs to treat skin cancer.