Enhancing quantitative structure-property relationship models by integrating complete molecular structure with deep learning

Abstract

Traditional quantitative structure-property relationship (QSPR) methods rely on molecular descriptors to quantify molecular structures and establish correlations with physical properties. In this study, we propose an approach that incorporates complete molecular structures to refine traditional QSPR methods and improve predictive accuracy. The supercritical properties used for modeling are collected from the literature. Molecular structures are optimized using density functional theory, from which molecular descriptors are derived. Both the structures and descriptors serve as inputs to the models developed in this work. Three models are constructed: a traditional artificial neural network model, a ResNet model, and a convolutional neural network (CNN)-enhanced model. Comparison with the JOBACK method shows that the CNN-enhanced model achieves higher predictive accuracy, whereas the ResNet model, which relies solely on molecular structures, suffers from pronounced overfitting.

Graphical abstract

Keywords

quantitative structure-property relationship / supercritical properties / artificial neural network / molecular structure / convolutional neural network

References

[1] Andrews T . On the continuity of the gaseous and liquid states of matter. Philosophical Transactions of the Royal Society of London, 1869, 159: 575–590

[2] Redlich O , Joseph N S K . On the thermodynamics of solutions. Chemical Reviews, 1949, 44(1): 233–244

[3] Peng D , Robinson D B . A new two-constant equation of state. Industrial & Engineering Chemistry Fundamentals, 1976, 15(1): 59–64

[4] Zhou L , Wang B , Jiang J , Pan Y , Wang Q . Quantitative structure-property relationship (QSPR) study for predicting gas-liquid critical temperatures of organic compounds. Thermochimica Acta, 2017, 655: 112–116

[5] Shi Z , Song F , Ji C , Xiao Y , Peng C , Liu H . QSPR models based on multiple linear regression and back-propagation artificial neural network for predicting electrical conductivity of ionic liquids. Industrial & Engineering Chemistry Research, 2024, 63(10): 4571–4584

[6] Liu X , Wang X , Yu M , Jia Q , Yan F , Wang Q . QSPR model to predict the speed of sound of ionic liquids as a function of variable temperature and pressure. Industrial & Engineering Chemistry Research, 2023, 62(17): 6772–6779

[7] Shi Y , Yu M , Liu J , Yan F , Luo Z H , Zhou Y N . Quantitative structure-property relationship model for predicting the propagation rate coefficient in free-radical polymerization. Macromolecules, 2022, 55(21): 9397–9410

[8] Ma Y , Gao Z , Shi P , Chen M , Wu S , Yang C , Wang J , Cheng J , Gong J . Machine learning-based solubility prediction and methodology evaluation of active pharmaceutical ingredients in industrial crystallization. Frontiers of Chemical Science and Engineering, 2022, 16(4): 523–535

[9] Espinosa G , Yaffe D , Arenas A , Cohen Y , Giralt F . A fuzzy ARTMAP-based quantitative structure-property relationship (QSPR) for predicting physical properties of organic compounds. Industrial & Engineering Chemistry Research, 2001, 40(12): 2757–2766

[10] Ghomisheh Z , Gorji A E , Sobati M A . Prediction of critical properties of sulfur-containing compounds: new QSPR models. Journal of Molecular Graphics & Modelling, 2020, 101: 107700

[11] Roubehie Fissa M , Lahiouel Y , Khaouane L , Hanini S . Development of QSPR-ANN models for the estimation of critical properties of pure hydrocarbons. Journal of Molecular Graphics & Modelling, 2023, 121: 108450

[12] Yan F , He W , Jia Q , Wang Q , Xia S , Ma P . Prediction of ionic liquids viscosity at variable temperatures and pressures. Chemical Engineering Science, 2018, 184: 134–140

[13] Liu X , Yu M , Jia Q , Yan F , Zhou Y N , Wang Q . Leave-one-ion-out cross-validation for assisting in developing robust QSPR models of ionic liquids. Journal of Molecular Liquids, 2023, 388: 122711

[14] Yan F , Shi Y , Wang Y , Jia Q , Wang Q , Xia S . QSPR models for the properties of ionic liquids at variable temperatures based on norm descriptors. Chemical Engineering Science, 2020, 217: 115540

[15] Egolf L M , Wessel M D , Jurs P C . Prediction of boiling points and critical temperatures of industrially important organic compounds from molecular structure. Journal of Chemical Information and Computer Sciences, 1994, 34(4): 947–956

[16] Turner B E , Costello C L , Jurs P C . Prediction of critical temperatures and pressures of industrially important organic compounds from molecular structure. Journal of Chemical Information and Computer Sciences, 1998, 38(4): 639–645

[17] Sobati M A , Abooali D . Molecular based models for estimation of critical properties of pure refrigerants: quantitative structure property relationship (QSPR) approach. Thermochimica Acta, 2015, 602: 53–62

[18] Gharagheizi F , Eslamimanesh A , Mohammadi A H , Richon D . Determination of critical properties and acentric factors of pure compounds using the artificial neural network group contribution algorithm. Journal of Chemical & Engineering Data, 2011, 56(5): 2460–2476

[19] Li Y , Sun Q , Zhu Z , Wen H , Jin S , Zhang X , Lei Z , Shen W . Multi-scale revolution of artificial intelligence in chemical industry. Frontiers of Chemical Science and Engineering, 2025, 19(7): 57

[20] Shen J , Yan M , Fang M , Gao X . Machine learning-based modeling approaches for estimating pyrolysis products of varied biomass and operating conditions. Bioresource Technology Reports, 2022, 20: 101285

[21] Lu L , Gao X , Dietiker J F , Shahnam M , Rogers W A . Machine learning accelerated discrete element modeling of granular flows. Chemical Engineering Science, 2021, 245: 116832

[22] Goyal A , Gai G , Cheng Z , Cunha J P , Zhu L , Wachs A . Flow past a random array of statistically homogeneously distributed stationary platonic polyhedrons: data analysis, probability maps, and deep learning models. International Journal of Multiphase Flow, 2024, 177: 104854

[23] Chew J W , Cocco R A . Application of machine learning methods to understand and predict circulating fluidized bed riser flow characteristics. Chemical Engineering Science, 2020, 217: 115503

[24] Ouyang B , Zhu L T , Su Y H , Luo Z H . A hybrid mesoscale closure combining CFD and deep learning for coarse-grid prediction of gas-particle flow dynamics. Chemical Engineering Science, 2022, 248: 117268

[25] Ouyang B , Zhu L T , Luo Z H . Machine learning for full spatiotemporal acceleration of gas-particle flow simulations. Powder Technology, 2022, 408: 117701

[26] Yaws C L . Chemical Properties Handbook: Physical, Thermodynamic, Environmental, Transport, Safety, and Health Related Properties for Organic and Inorganic Chemicals. New York: McGraw Hill, 1999,

[27] Liu Q , Jiang Y , Zhang L , Du J . A computational toolbox for molecular property prediction based on quantum mechanics and quantitative structure-property relationship. Frontiers of Chemical Science and Engineering, 2022, 16(2): 152–167

[28] Shi Y , Wang J , Wang Q , Jia Q , Yan F , Luo Z H , Zhou Y N . Supervised machine learning algorithms for predicting rate constants of ozone reaction with micropollutants. Industrial & Engineering Chemistry Research, 2022, 61(24): 8359–8367

[29] Shi Y , Li J J , Wang Q , Jia Q , Yan F , Luo Z H , Zhou Y N . Computer-aided estimation of kinetic rate constant for degradation of volatile organic compounds by hydroxyl radical: an improved model using quantum chemical and norm descriptors. Chemical Engineering Science, 2022, 248: 117244

[30] Shi Y , Han J P , Yan F , Luo Z H , Zhou Y N . In silico prediction of the temperature-dependent decomposition rate coefficients of symmetrical azo compounds. Industrial & Engineering Chemistry Research, 2023, 62(43): 17461–17472

[31] Lu T , Chen F . Multiwfn: a multifunctional wavefunction analyzer. Journal of Computational Chemistry, 2012, 33(5): 580–592

[32] Ouyang B , Zhu L , Luo Z . Interpretable machine learning analysis and automated modeling to simulate fluid-particle flows. Particuology, 2023, 80: 42–52

[33] Cao D , Chen Y , Chen J , Zhang H , Yuan Z . An improved algorithm for the maximal information coefficient and its application. Royal Society Open Science, 2021, 8(2): 201424

[34] Xie Z , Gu X , Shen Y . A machine learning study of predicting mixing and segregation behaviors in a bidisperse solid-liquid fluidized bed. Industrial & Engineering Chemistry Research, 2022, 61(24): 8551–8565

[35] Li W , Gao X , Hao Z , Sun R . Using deep learning for precipitation forecasting based on spatio-temporal information: a case study. Climate Dynamics, 2022, 58(1): 443–457

[36] Owoyele O , Pal P , Vidal Torreira A . An automated machine learning-genetic algorithm framework with active learning for design optimization. Journal of Energy Resources Technology, 2021, 143(8): 082305

[37] Farooq H , Saeed A , Akhtar I , Bangash Z . Neural network-based model reduction of hydrodynamics forces on an airfoil. Fluids, 2021, 6(9): 332

[38] Jin HSong QHu X. Auto-Keras: an efficient neural architecture search system. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York: Association for Computing Machinery, 2019, 1946–1956

[39] Zhang D , Ouyang B , Luo Z H . Identification of gas-solid flow regimes using convolutional neural network techniques. Powder Technology, 2024, 442: 119848

[40] Kim Y , Jung H , Kumar S , Paton R S , Kim S . Designing solvent systems using self-evolving solubility databases and graph neural networks. Chemical Science, 2024, 15(3): 923–939

[41] Hussain E , Tahir M H , Dalal A A , Naeem S , Islam H E A . Polymer design with enhanced crystallization tendency aided by machine learning. Physica B: Condensed Matter, 2024, 694: 416437

[42] Yao X , Wang Y , Zhang X , Zhang R , Liu M , Hu Z , Fan B . Radial basis function neural network-based QSPR for the prediction of critical temperature. Chemometrics and Intelligent Laboratory Systems, 2002, 62(2): 217–225

[43] Kim Y , Cho J , Jung H , Meyer L E , Fioroni G M , Stubbs C D , Jeong K , McCormick R L , John P C S , Kim S . Designing green chemicals by predicting vaporization properties using explainable graph attention networks. Green Chemistry, 2024, 26(19): 10247–10264

[44] Banchero M , Manna L . Comparison between multi-linear- and radial-basis-function-neural-network-based QSPR models for the prediction of the critical temperature, critical pressure, and acentric factor of organic compounds. Molecules, 2018, 23(6): 1379

[45] Aouichaoui A R , Mansouri S S , Abildskov J , Sin G . Uncertainty estimation in deep learning-based property models: graph neural networks applied to the critical properties. AIChE Journal, 2022, 68(6): e17696

RIGHTS & PERMISSIONS

Higher Education Press