🔢 mathematics

Graph-Theoretic Computation of Degree-Based Topological Indices for Hepatocellular Carcinoma Therapeutics: A Mathematical QSPR Framework for Physicochemical Property Prediction

This study establishes a rigorous graph-theoretic framework for predicting the physicochemical properties of five Hepatocellular Carcinoma therapeutics by demonstrating that six degree-based topological indices, particularly the Yemen Index, exhibit statistically significant correlations with Surface Tension through linear regression analysis.

Original authors: S. Ranjitham, O. V. Shanmuga Sundaram, G. Priyadharsini, A. Saravanan

Published 2026-06-26

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: S. Ranjitham, O. V. Shanmuga Sundaram, G. Priyadharsini, A. Saravanan

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Turning Molecules into Math

Imagine you have a complex Lego structure (a drug molecule) that fights liver cancer. Usually, to understand how this structure behaves in the real world—like how sticky it is or how hot it needs to get to boil—you have to build it in a lab and run expensive, time-consuming experiments.

This paper proposes a shortcut. Instead of building the Lego set, the researchers looked at the blueprint (the molecular graph) and used a special set of mathematical rulers (topological indices) to predict how the drug would behave.

The Cast of Characters

The study focused on five specific drugs used to treat Hepatocellular Carcinoma (a type of liver cancer):

Futibatinib
Ipilimumab
Pembrolizumab
Pemigatinib
Regorafenib

The researchers treated each drug like a map. In this map:

Atoms are the cities (dots/vertices).
Chemical bonds are the roads connecting them (lines/edges).

The Six "Mathematical Rulers"

To measure the "shape" of these drug maps, the team used six different mathematical formulas. Think of these as six different ways to weigh a suitcase:

First & Second Zagreb Indices: Basic ways to count how busy the intersections are.
Forgotten Index: A version that gives extra weight to very busy intersections.
Yemen Index: A more sensitive version that really notices the busiest hubs.
Degree Index: An extreme version that blows up the numbers for the busiest hubs (like squaring or cubing the count).
Sum-Connectivity Index: Another variation that looks at how roads connect.

The Experiment: What Did They Measure?

The team tried to see if these six "rulers" could predict five real-world physical properties of the drugs:

Boiling Point: How hot before it turns to gas?
Flash Point: How hot before it catches fire?
Polarizability: How easily the molecule's electrons wiggle when pushed.
Surface Tension: How "sticky" or cohesive the liquid is (like how water beads up on a leaf).
Molar Volume: How much space the molecules take up.

The Results: The "Sticky" Surprise

Here is the main finding, explained simply:

1. The "Sticky" Winner (Surface Tension)
The mathematical rulers were amazing at predicting Surface Tension.

The Analogy: Imagine trying to guess how sticky a piece of tape is just by looking at the pattern of its fibers. The researchers found that if you use the right math formula (specifically the Yemen Index), you can guess the stickiness of these cancer drugs with 97.8% accuracy.
Why? The paper suggests that surface tension is all about how molecules hold hands with their neighbors. The mathematical rulers are very good at counting how many "hands" (connections) a molecule has, which directly relates to how sticky it is.

2. The "Misses" (Boiling Point, Flash Point, etc.)
The rulers failed to accurately predict the other four properties (Boiling Point, Flash Point, Polarizability, Molar Volume).

The Analogy: It's like trying to guess the color of a car just by counting its wheels. You can count the wheels perfectly (the math works), but the number of wheels doesn't tell you if the car is red or blue.
Why? The paper explains that things like boiling point depend on complex, invisible factors (like quantum mechanics and specific 3D shapes) that these simple "connection-counting" math formulas can't see. Also, the study only had 5 drugs to test, which isn't enough data to make a perfect guess for these harder properties.

The "Magic" Formula

Out of all the math tools they used, the Yemen Index was the champion.

The Result: It created a perfect straight-line relationship with Surface Tension.
The Takeaway: If you know the structure of a liver cancer drug, you can use this specific math formula to tell you exactly how "cohesive" or "sticky" the liquid drug will be.

What the Paper Doesn't Say

It is important to stick to what the authors actually claimed:

They did not say this method will cure cancer.
They did not say this will replace all lab experiments forever.
They did not claim it works for every type of drug, only that it worked very well for these five specific liver cancer drugs regarding surface tension.

Summary

This paper is like a detective story where the detective (the mathematician) solves a mystery (predicting drug behavior) using only a blueprint. They discovered that while the blueprint can't tell you everything (like the boiling point), it is a super-accurate crystal ball for predicting how "sticky" the drug liquid will be. This could help scientists design better drug formulas faster, without needing to mix chemicals in a lab for every single guess.

Technical Summary: Graph-Theoretic Computation of Degree-Based Topological Indices for Hepatocellular Carcinoma Therapeutics

Problem Statement
The experimental determination of physicochemical properties (such as boiling point, flash point, polarizability, surface tension, and molar volume) for drug candidates is resource-intensive and difficult to scale for large compound libraries. While Quantitative Structure–Property Relationship (QSPR) frameworks offer a mathematical alternative by modeling molecules as graphs, there is a specific gap in the literature regarding the application of a comprehensive battery of degree-based topological indices to Hepatocellular Carcinoma (HCC) therapeutics. This study addresses the need for a computationally efficient, graph-theoretic framework to predict the physicochemical behavior of five clinically approved HCC drugs directly from their molecular structures.

Methodology
The research employs Chemical Graph Theory (CGT) to model five HCC drugs: Futibatinib, Ipilimumab, Pembrolizumab, Pemigatinib, and Regorafenib.

Graph Construction: Each molecule is represented as a simple, connected molecular graph $G(V, E)$ , where vertices ( $V$ ) represent non-hydrogen heavy atoms and edges ( $E$ ) represent covalent bonds. Hydrogen atoms are suppressed following standard conventions.
Topological Indices: Six degree-based topological indices were computed for each molecular graph. These indices vary in their sensitivity to vertex degrees ( $d_v$ $d_{v}$ ) through different power summations:
- First Zagreb Index ( $M_1$ ): Sum of $(d_u + d_v)$ .
- Second Zagreb Index ( $M_2$ ): Sum of $(d_u \times d_v)$ .
- Forgotten Index ( $F$ ): Sum of $(d_u^2 + d_v^2)$ .
- Yemen Index ( $Y$ ): Sum of $(d_u^3 + d_v^3)$ .
- Degree Index ( $D$ ): Sum of $(d_u^6 + d_v^6)$ .
- Sum-Connectivity Index ( $S$ ): Sum of $(d_u^4 + d_v^4)$ .
Data and Analysis: Experimental physicochemical data (Boiling Point, Flash Point, Polarizability, Surface Tension, Molar Volume) were retrieved from the ChemSpider database. Linear regression models ( $T = \alpha + \beta t_i$ ) were constructed for all 30 index–property combinations using SPSS (v25.0). Model performance was evaluated using Pearson correlation coefficients ( $r$ ), coefficients of determination ( $R^2$ ), and statistical significance ( $p < 0.05$ ).

Key Results
The study yielded distinct findings regarding the predictive capability of the indices across different properties:

Surface Tension (ST) Dominance: Surface Tension was the only physicochemical property that demonstrated statistically significant correlations ( $p < 0.05$ ) with the topological indices. Five of the six indices produced significant QSPR models for ST, with $R^2$ values ranging from 0.907 to 0.978.
Index Performance Hierarchy: The Yemen Index ( $Y(G)$ ) achieved the highest predictive fidelity ( $R^2 = 0.978$ , $p = 0.011$ ), followed closely by the Forgotten Index ( $F(G)$ , $R^2 = 0.967$ ) and the Sum-Connectivity Index ( $S(G)$ , $R^2 = 0.965$ ). The First and Second Zagreb indices also showed significance but with lower $R^2$ values.
Non-Significant Properties: Models for Boiling Point, Flash Point, Polarizability, and Molar Volume failed to achieve statistical significance. The authors attribute this to the small sample size (reduced to $N=4$ for Boiling and Flash points due to missing data for Pemigatinib) and the multifactorial nature of these properties, which rely on electronic and steric factors not fully captured by degree-based descriptors alone.
The "Degree Index" Anomaly: The Degree Index ( $D(G)$ ), which uses a sixth-power summation, showed a high correlation ( $r=0.933$ ) with Surface Tension but failed to reach statistical significance ( $p=0.067$ ). The authors explain this as an artifact of the extreme dynamic range of $D(G)$ values (spanning nearly two orders of magnitude), which inflates variance in small-sample linear regression.

Significance and Claims
The paper establishes a mathematically tractable framework for predicting the surface-active physicochemical behavior of HCC therapeutics based solely on molecular graph structure.

Mechanistic Insight: The authors propose that the success of degree-based indices (particularly cubic and quadratic powers like $Y(G)$ and $F(G)$ ) in predicting Surface Tension arises from the alignment between these indices and the density of high-connectivity vertices (e.g., aromatic ring junctions). These structural features correlate with the London dispersion interactions that govern liquid-phase cohesion.
Practical Application: The study claims that the ability to computationally predict Surface Tension without experimental measurement offers direct implications for pharmaceutical formulation, specifically in optimizing wettability, spreading, and emulsification for HCC drug delivery systems.
Scope Limitations: The authors remain modest regarding the generalizability of the results to other properties (BP, FP, P, MV), noting that degree-only descriptors are insufficient for these parameters without the inclusion of 3D, electronic, or distance-based descriptors. They suggest that future work should expand the compound set and integrate non-linear machine learning approaches to address these limitations.