Shampa Raghunathan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Shampa Raghunathan has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Markus Meuwly
University of Basel Klingelbergstrasse 80
3 shared publications Mathias Nest
Theoretische Chemie, TU München, Lichtenbergstr. 4, 85747 Garching, Germany
3 shared publications- Udaykumar Deva Priyakumar
Center for Computational Natural Sciences and Bioinformatics, International Institute of Information Technology, Hyderabad, India
3 shared publications - Mathias Nest
Theoretische Chemie, TU München, Lichtenbergstr. 4, 85747 Garching, Germany
2 shared publications Tanashree Jaganade
2 shared publicationsAditya Chattopadhyay
1 shared publication
Latest Publications
- 30 December 2025
- Published by IOP Publishing in Journal of Physics: Condensed Matter
- Vol. 38 (1) , 013002
Machine-learned potentials for solvation modeling
Solvent environments play a central role in determining molecular structure, energetics, reactivity, and interfacial phenomena. However, modeling solvation from first principles remains difficult due to the complex interplay of interactions and unfavorable computational scaling of first-principles treatment with system size. Machine-learned potentials (MLPs) have recently emerged as efficient surrogates for quantum chemistry methods, offering first-principles accuracy at greatly reduced computational cost. MLPs approximate the underlying potential energy surface, enabling efficient computation of energies and forces in solvated systems. They are also capable of accounting for effects such as hydrogen bonding, long-range polarization, and conformational changes. This review surveys the development and application of MLPs in solvation modeling. We summarize the theoretical basis of MLP-based energy and force predictions and present a classification of MLPs based on training targets, model types, and design choices related to architectures, descriptors, and training protocols. Integration into established solvation workflows is discussed, with case studies spanning small molecules, interfaces, and reactive systems. We conclude by outlining open challenges and future directions toward transferable, robust, and physically grounded MLPs for solvation-aware atomistic modeling.
- 30 December 2025
- Published by IOP Publishing in Journal of Physics: Condensed Matter
- Vol. 38 (1) , 013002
Machine-learned potentials for solvation modeling
Solvent environments play a central role in determining molecular structure, energetics, reactivity, and interfacial phenomena. However, modeling solvation from first principles remains difficult due to the complex interplay of interactions and unfavorable computational scaling of first-principles treatment with system size. Machine-learned potentials (MLPs) have recently emerged as efficient surrogates for quantum chemistry methods, offering first-principles accuracy at greatly reduced computational cost. MLPs approximate the underlying potential energy surface, enabling efficient computation of energies and forces in solvated systems. They are also capable of accounting for effects such as hydrogen bonding, long-range polarization, and conformational changes. This review surveys the development and application of MLPs in solvation modeling. We summarize the theoretical basis of MLP-based energy and force predictions and present a classification of MLPs based on training targets, model types, and design choices related to architectures, descriptors, and training protocols. Integration into established solvation workflows is discussed, with case studies spanning small molecules, interfaces, and reactive systems. We conclude by outlining open challenges and future directions toward transferable, robust, and physically grounded MLPs for solvation-aware atomistic modeling.
- 14 May 2025
- Published by Wiley in Journal of Computational Chemistry
- Vol. 46 (13) , e70121
Graph Neural Network for 3‐Dimensional Structures Including Dihedral Angles for Molecular Property Prediction
The prediction of molecular properties using graph neural network (GNN)‐ based approaches has attracted great attention in recent years. Topological molecular graphs are commonly used for representing molecules in machine learning (ML). However, the challenge is to utilize the complete geometry information, such as, bonds, angles, and dihedral angles while processing a molecular graph. In this work, we present predictive GNN accounting three‐dimensional molecular structures including the dihedral angles (GNN3Dihed) in a systematic manner. Additionally, we demonstrate that the usage of autoencoders to generate latent space embeddings for usually sparse atomic and bond vectors reduces the number of parameters in the message passing stage while not reducing performance. We compare the performance of GNN3Dihed with state‐of‐the‐art baselines on several tasks (regression and classification), for example, solubility prediction, toxicity prediction, binding affinity, and quantum mechanical property prediction, and showed that the present architecture often outperforms other models—demonstrating the importance of 3D structural information for ML in chemistry.
- 14 May 2025
- Published by Wiley in Journal of Computational Chemistry
- Vol. 46 (13) , e70121
Graph Neural Network for 3‐Dimensional Structures Including Dihedral Angles for Molecular Property Prediction
The prediction of molecular properties using graph neural network (GNN)‐ based approaches has attracted great attention in recent years. Topological molecular graphs are commonly used for representing molecules in machine learning (ML). However, the challenge is to utilize the complete geometry information, such as, bonds, angles, and dihedral angles while processing a molecular graph. In this work, we present predictive GNN accounting three‐dimensional molecular structures including the dihedral angles (GNN3Dihed) in a systematic manner. Additionally, we demonstrate that the usage of autoencoders to generate latent space embeddings for usually sparse atomic and bond vectors reduces the number of parameters in the message passing stage while not reducing performance. We compare the performance of GNN3Dihed with state‐of‐the‐art baselines on several tasks (regression and classification), for example, solubility prediction, toxicity prediction, binding affinity, and quantum mechanical property prediction, and showed that the present architecture often outperforms other models—demonstrating the importance of 3D structural information for ML in chemistry.
- Open Access
- 13 November 2024
- Published by American Chemical Society (ACS) in ChemRxiv
Graph neural network for 3-dimensional structures including dihedral angles for molecular property prediction
The prediction of molecular properties using graph neural network (GNN) based approaches has attracted great attention in recent years. Topological molecular graphs are commonly used for representing molecules in machine learning (ML). However, the challenge is to utilize the complete geometry information, like, bonds, angles and dihedral angles while processing a molecular graph. In this work we present predictive GNN accounting three-dimensional molecular structures including the dihedral angles (GNN3Dihed) in a systematic manner. Additionally, we demonstrate that the usage of autoencoders to generate latent space embeddings for usually sparse atomic and bond vectors reduces the number of parameters in the message passing stage while not reducing performance. We compare the performance of GNN3Dihed with state-of-the-art baselines on several tasks (regression and classification), e.g., solubility prediction, toxicity prediction, binding affinity, and quantum mechanical property prediction, and showed that the present architecture often outperforms other models–demonstrating the importance of 3D structural information for ML in chemistry.
- 13 November 2024
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessGraph neural network for 3-dimensional structures including dihedral angles for molecular property prediction
The prediction of molecular properties using graph neural network (GNN) based approaches has attracted great attention in recent years. Topological molecular graphs are commonly used for representing molecules in machine learning (ML). However, the challenge is to utilize the complete geometry information, like, bonds, angles and dihedral angles while processing a molecular graph. In this work we present predictive GNN accounting three-dimensional molecular structures including the dihedral angles (GNN3Dihed) in a systematic manner. Additionally, we demonstrate that the usage of autoencoders to generate latent space embeddings for usually sparse atomic and bond vectors reduces the number of parameters in the message passing stage while not reducing performance. We compare the performance of GNN3Dihed with state-of-the-art baselines on several tasks (regression and classification), e.g., solubility prediction, toxicity prediction, binding affinity, and quantum mechanical property prediction, and showed that the present architecture often outperforms other models–demonstrating the importance of 3D structural information for ML in chemistry.
- Open Access
- 5 August 2024
- Published by Royal Society of Chemistry (RSC) in RSC Advances
- Vol. 14 (34) , 25031-25041
Solvent accessible surface area-assessed molecular basis of osmolyte-induced protein stability
A generic model for solvent-modulated protein folding is developed based on statistical mechanics and the thermodynamic transfer free energy model by considering the microenvironment polarity of the interacting co(solvent)–protein system.
- 5 August 2024
- Published by Royal Society of Chemistry (RSC) in RSC Advances
- Vol. 14 (34) , 25031-25041
Open AccessSolvent accessible surface area-assessed molecular basis of osmolyte-induced protein stability
A generic model for solvent-modulated protein folding is developed based on statistical mechanics and the thermodynamic transfer free energy model by considering the microenvironment polarity of the interacting co(solvent)–protein system.
- Open Access
- 13 July 2023
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 4 (3) , 035006
Intramolecular proton transfer reaction dynamics using machine-learned ab initio potential energy surfaces
Hydrogen bonding interactions, which are central to various physicochemical processes, are investigated in the present study using ab initio-based machine learning potential energy surfaces. Abnormally strong intramolecular O–H⋯O hydrogen bonds, occurring in β-diketone enols of malonaldehyde and its derivatives, with substituents ranging from various electron-withdrawing to electron-donating functional groups, are studied. Machine learning force fields were constructed using a kernel-based force learning model employing ab initio molecular dynamics reference data. These models were used for molecular dynamics simulations at finite temperature, and dynamical properties were determined by computing proton transfer free-energy surfaces. The chemical systems studied here show progression toward barrier-less proton transfer events at an accuracy of correlated electronic structure methods. Markov state models of the conformational states indicate shorter intramolecular hydrogen bonds exhibiting higher proton transfer rates. We demonstrate how functional group substitution can modulate the strength of intramolecular hydrogen bonds by studying the thermodynamic and kinetic properties.
- 13 July 2023
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 4 (3) , 035006
Open AccessIntramolecular proton transfer reaction dynamics using machine-learned ab initio potential energy surfaces
Hydrogen bonding interactions, which are central to various physicochemical processes, are investigated in the present study using ab initio-based machine learning potential energy surfaces. Abnormally strong intramolecular O–H⋯O hydrogen bonds, occurring in β-diketone enols of malonaldehyde and its derivatives, with substituents ranging from various electron-withdrawing to electron-donating functional groups, are studied. Machine learning force fields were constructed using a kernel-based force learning model employing ab initio molecular dynamics reference data. These models were used for molecular dynamics simulations at finite temperature, and dynamical properties were determined by computing proton transfer free-energy surfaces. The chemical systems studied here show progression toward barrier-less proton transfer events at an accuracy of correlated electronic structure methods. Markov state models of the conformational states indicate shorter intramolecular hydrogen bonds exhibiting higher proton transfer rates. We demonstrate how functional group substitution can modulate the strength of intramolecular hydrogen bonds by studying the thermodynamic and kinetic properties.
- Open Access
- 17 August 2022
- Published by American Chemical Society (ACS) in ChemRxiv
Solvent accessible surface area-assessed molecular basis of osmolyte-induced protein stability
In the solvent modulated protein folding, under certain physiological conditions, an equilibrium exists between the unfolded and folded states of the protein without any need to break or make a covalent bond. Here, interactions between various protein groups (peptides) and solvent molecules play a major role in determining the directionality of the chemical reaction. However, quantitative determination of such interactions lacks any unified theory. To this end, a solvation model is developed based on statistical mechanics, and the thermodynamic transfer free energy model. According to this model, polarity and the fractional solvent accessible surface areas contribute to the interaction energies. The current work is inspired by Street et al., Proc. Natl. Acad. Sci. USA 2006, 103, 13997; that was specifically developed for predicting transfer free energies for protein backbone. The present model further develops the method in order to include various orientations of participating interactant surfaces of suitable areas. The model is evaluated for amino acid side chains besides backbone. As model systems, we consider naturally occurring amino acid residues solvated in ten different osmolytes which are small organic compounds known to modulate protein stability. The present model is able to predict the correct trend of the osmolyte-peptide interactions ranging from the stabilizing to destabilizing. The versatility of this model facilitates calculating any solute-solvent interaction energies.
- 17 August 2022
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessSolvent accessible surface area-assessed molecular basis of osmolyte-induced protein stability
In the solvent modulated protein folding, under certain physiological conditions, an equilibrium exists between the unfolded and folded states of the protein without any need to break or make a covalent bond. Here, interactions between various protein groups (peptides) and solvent molecules play a major role in determining the directionality of the chemical reaction. However, quantitative determination of such interactions lacks any unified theory. To this end, a solvation model is developed based on statistical mechanics, and the thermodynamic transfer free energy model. According to this model, polarity and the fractional solvent accessible surface areas contribute to the interaction energies. The current work is inspired by Street et al., Proc. Natl. Acad. Sci. USA 2006, 103, 13997; that was specifically developed for predicting transfer free energies for protein backbone. The present model further develops the method in order to include various orientations of participating interactant surfaces of suitable areas. The model is evaluated for amino acid side chains besides backbone. As model systems, we consider naturally occurring amino acid residues solvated in ten different osmolytes which are small organic compounds known to modulate protein stability. The present model is able to predict the correct trend of the osmolyte-peptide interactions ranging from the stabilizing to destabilizing. The versatility of this model facilitates calculating any solute-solvent interaction energies.
- Open Access
- 27 December 2021
- Published by Wiley in International Journal of Quantum Chemistry
- Vol. 122 (7) , e26870
Molecular representations for machine learning applications in chemistry
Machine learning (ML) methods enable computers to address problems by learning from existing data. Such applications are becoming commonplace in molecular sciences. Interest in applying ML techniques across chemical compound space, from predicting properties to designing molecules and materials is in the surge. Especially, ML models have started to accelerate computational chemistry, and are often as accurate as state‐of‐the‐art electronic/atomistic models. Being an integral part of the ML architecture, representation of a molecular entity, uniquely encoded, plays a crucial role to what extent an ML model would be accurately predicting the desired property. This review aims to demonstrate a hierarchy of representations which has been introduced, to capture all degrees of freedom of a molecule or an atom the best, to map the quantum mechanical properties. We discuss their diverse applications how they have been instrumental in harnessing the growing field of ML accelerated computational modeling.
- 27 December 2021
- Published by Wiley in International Journal of Quantum Chemistry
- Vol. 122 (7) , e26870
Open AccessMolecular representations for machine learning applications in chemistry
Machine learning (ML) methods enable computers to address problems by learning from existing data. Such applications are becoming commonplace in molecular sciences. Interest in applying ML techniques across chemical compound space, from predicting properties to designing molecules and materials is in the surge. Especially, ML models have started to accelerate computational chemistry, and are often as accurate as state‐of‐the‐art electronic/atomistic models. Being an integral part of the ML architecture, representation of a molecular entity, uniquely encoded, plays a crucial role to what extent an ML model would be accurately predicting the desired property. This review aims to demonstrate a hierarchy of representations which has been introduced, to capture all degrees of freedom of a molecule or an atom the best, to map the quantum mechanical properties. We discuss their diverse applications how they have been instrumental in harnessing the growing field of ML accelerated computational modeling.
- Open Access
- 5 July 2021
- Published by Cambridge University Press (CUP) in ChemRxiv
Intramolecular proton transfer reaction dynamics using machine learned ab initio potential energy surfaces
Hydrogen bonding interactions central to various physicochemical processes are investigated in the present study using ab initio-based machine learning potential energy surfaces. Abnormally strong intramolecular O-H...O hydrogen bonds occurring in beta-diketone enols of malonaldehyde, and its derivatives with substituents ranging from various electron-withdrawing to electron-donating functional groups are studied. Machine learning force fields were constructed by using a kernel-based force learning model employing ab initio molecular dynamics reference data. These models were used for molecular dynamics simulations at finite temperature, and dynamical properties were determined by computing their proton transfer free energy surfaces. Chemical systems studied here show progression towards forming barrierless proton transfer events at an accuracy of highly correlated electronic structure methods. Markov state models of the conformational states indicate shorter intramolecular hydrogen bonds exhibiting higher proton transfer rates. We demonstrate how functional group substitution can modulate the strength of intramolecular hydrogen bonds by studying their thermodynamic and kinetic properties.
- 5 July 2021
- Published by Cambridge University Press (CUP) in ChemRxiv
Open AccessIntramolecular proton transfer reaction dynamics using machine learned ab initio potential energy surfaces
Hydrogen bonding interactions central to various physicochemical processes are investigated in the present study using ab initio-based machine learning potential energy surfaces. Abnormally strong intramolecular O-H...O hydrogen bonds occurring in beta-diketone enols of malonaldehyde, and its derivatives with substituents ranging from various electron-withdrawing to electron-donating functional groups are studied. Machine learning force fields were constructed by using a kernel-based force learning model employing ab initio molecular dynamics reference data. These models were used for molecular dynamics simulations at finite temperature, and dynamical properties were determined by computing their proton transfer free energy surfaces. Chemical systems studied here show progression towards forming barrierless proton transfer events at an accuracy of highly correlated electronic structure methods. Markov state models of the conformational states indicate shorter intramolecular hydrogen bonds exhibiting higher proton transfer rates. We demonstrate how functional group substitution can modulate the strength of intramolecular hydrogen bonds by studying their thermodynamic and kinetic properties.
- Open Access
- 29 June 2021
- Published by Cambridge University Press (CUP) in ChemRxiv
Intramolecular proton transfer reaction dynamics using machine learned ab initio potential energy surfaces
Hydrogen bonding interactions central to various physicochemical processes are investigated in the present study using ab initio-based machine learning potential energy surfaces. Abnormally strong intramolecular O-H...O hydrogen bonds occurring in beta-diketone enols of malonaldehyde, and its derivatives with substituents ranging from various electron-withdrawing to electron-donating functional groups are studied. Machine learning force fields were constructed by using a kernel-based force learning model employing ab initio molecular dynamics reference data. These models were used for molecular dynamics simulations at finite temperature, and dynamical properties were determined by computing their proton transfer free energy surfaces. Chemical systems studied here show progression towards forming barrierless proton transfer events at an accuracy of highly correlated electronic structure methods. Markov state models of the conformational states indicate shorter intramolecular hydrogen bonds exhibiting higher proton transfer rates. We demonstrate how functional group substitution can modulate the strength of intramolecular hydrogen bonds by studying their thermodynamic and kinetic properties.
- 29 June 2021
- Published by Cambridge University Press (CUP) in ChemRxiv
Open AccessIntramolecular proton transfer reaction dynamics using machine learned ab initio potential energy surfaces
Hydrogen bonding interactions central to various physicochemical processes are investigated in the present study using ab initio-based machine learning potential energy surfaces. Abnormally strong intramolecular O-H...O hydrogen bonds occurring in beta-diketone enols of malonaldehyde, and its derivatives with substituents ranging from various electron-withdrawing to electron-donating functional groups are studied. Machine learning force fields were constructed by using a kernel-based force learning model employing ab initio molecular dynamics reference data. These models were used for molecular dynamics simulations at finite temperature, and dynamical properties were determined by computing their proton transfer free energy surfaces. Chemical systems studied here show progression towards forming barrierless proton transfer events at an accuracy of highly correlated electronic structure methods. Markov state models of the conformational states indicate shorter intramolecular hydrogen bonds exhibiting higher proton transfer rates. We demonstrate how functional group substitution can modulate the strength of intramolecular hydrogen bonds by studying their thermodynamic and kinetic properties.
- Open Access
- 21 May 2021
- Published by MDPI AG in International Journal of Molecular Sciences
- Vol. 22 (11) , 5411
Multiscale Modeling of Wobble to Watson–Crick-Like Guanine–Uracil Tautomerization Pathways in RNA
Energetically unfavorable Watson–Crick (WC)-like tautomeric forms of nucleobases are known to introduce spontaneous mutations, and contribute to replication, transcription, and translation errors. Recent NMR relaxation dispersion techniques were able to show that wobble (w) G•U mispair exists in equilibrium with the short-lived, low-population WC-like enolic tautomers. Presently, we have investigated the wG•U → WC-like enolic reaction pathway using various theoretical methods: quantum mechanics (QM), molecular dynamics (MD), and combined quantum mechanics/molecular mechanics (QM/MM). The previous studies on QM gas phase calculations were inconsistent with experimental data. We have also explored the environmental effects on the reaction energies by adding explicit water. While the QM-profile clearly becomes endoergic in the presence of water, the QM/MM-profile remains consistently endoergic in the presence and absence of water. Hence, by including microsolvation and QM/MM calculations, the experimental data can be explained. For the G•U→ G•U pathway, the latter appears to be energetically more favorable throughout all computational models. This study can be considered as a benchmark of various computational models of wG•U to WC-like tautomerization pathways with and without the environmental effects, and may contribute on further studies of other mispairs as well.- 21 May 2021
- Published by MDPI AG in International Journal of Molecular Sciences
- Vol. 22 (11) , 5411
Open AccessMultiscale Modeling of Wobble to Watson–Crick-Like Guanine–Uracil Tautomerization Pathways in RNA
Energetically unfavorable Watson–Crick (WC)-like tautomeric forms of nucleobases are known to introduce spontaneous mutations, and contribute to replication, transcription, and translation errors. Recent NMR relaxation dispersion techniques were able to show that wobble (w) G•U mispair exists in equilibrium with the short-lived, low-population WC-like enolic tautomers. Presently, we have investigated the wG•U → WC-like enolic reaction pathway using various theoretical methods: quantum mechanics (QM), molecular dynamics (MD), and combined quantum mechanics/molecular mechanics (QM/MM). The previous studies on QM gas phase calculations were inconsistent with experimental data. We have also explored the environmental effects on the reaction energies by adding explicit water. While the QM-profile clearly becomes endoergic in the presence of water, the QM/MM-profile remains consistently endoergic in the presence and absence of water. Hence, by including microsolvation and QM/MM calculations, the experimental data can be explained. For the G•U→ G•U pathway, the latter appears to be energetically more favorable throughout all computational models. This study can be considered as a benchmark of various computational models of wG•U to WC-like tautomerization pathways with and without the environmental effects, and may contribute on further studies of other mispairs as well.