SSSergio Suárez-Dou

Sergio Suárez-Dou

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Sergio Suárez-Dou has a total of 58 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Explainable ML force-field for evaluating protein–ligand binding energy using SO3LR

    by Hamza Agha, Sergio SuárEz-Dou, Adil Kabylda, Alexandre Tkatchenko, Andrea Volkamer

    Accurate estimation of protein-ligand binding affinities remains central to structure-based drug design. Classical force fields neglect important quantum-mechanical effects, while quantum mechanical and semi-empirical methods remain impractical for large biomolecular systems. Machine-learned force fields (MLFFs) offer an alternative, but their performance may be inflated by training on protein-ligand complexes. Here, we introduce SO3LR-SF, a physics-based scoring framework that can be instantiated with any of the four pretrained SO3LR MLFF models, combining an equivariant message-passing network for semi-local interactions with physically motivated terms for short-range repulsion, electrostatics, and long-range dispersion. On PLA15 (15 active-site models), SO3LR-SF attains a relative interaction error of 2.1-5.8% against DLPNO-CCSD(T) references depending on the underlying SO3LR model, with SO3LR-T-M reaching 2.1%, the lowest of all tested methods. On the FEP benchmark (8 targets, 264 ligands), all SO3LR-SF variants achieve an average Spearman correlation of around 0.50, on par with the best semi-empirical quantum-mechanical method tested (0.47, GFN-FF) and with MMGBSA (0.44), and surpassing Glide (0.27). On the Wang dataset (8 targets, 199 ligands), the SO3LR-SF models reach around 0.59, matching the 0.59-0.62 range of molecular dynamics-based free-energy methods at a fraction of their cost. Each complex is scored in 3 s to 47 s on 12 CPU cores. An 8 Å trimming protocol reduces runtime threefold without loss of accuracy, and a restrained geometry optimization recovers ranking accuracy for targets with strained input structures. A multi-level explainability framework adds energy decomposition, per-atom energy contributions reported either for ligand or as 2D protein-ligand interaction maps, and 3D binding site visualization. We found that the polar SASA ratio of the binding pocket is predictive of electrostatic scoring performance, enabling estimation of ligand-ranking accuracy before scoring (r = 0.74-0.81 across models). Together, SO3LR-SF presents a fast, interpretable, and competitive scoring function for drug discovery that requires no experimental binding-affinity data for training.

  • Open Access

    MARS: Machine-Learned Force Field Framework for Automated Conformational Sampling, Vibrational Spectroscopy, and Microsolvation

    by Sergio SuárEz-Dou, Miguel Gallegos, Florian N. BrünIg, Alexandre Tkatchenko

    We present MARS, an open-source Python package for automated conformational ensemble generation, infrared spectrum prediction, and explicit microsolvation of molecular systems. Built on JAX for GPU acceleration and automatic differentiation, MARS is carried out with machine-learned force fields (MLFFs), including SO3LR and MACE, with a four-phase conformational sampling workflow combining metadynamics, multitemperature molecular dynamics, and genetic algorithms. The package provides automated vibrational mode classification using Wilson B-matrix projections and functional group identification via graph-based pattern matching. We benchmark MARS on chemically diverse systems, including alkanes, halogenated aromatics, amino acids, peptides, drug-like molecules, phosphamides, host–guest complexes, and microsolvated biomolecular models. Semiempirical quantum-chemical methods have until now provided the practical state of the art for automated conformational sampling. Our benchmarks show that modern MLFFs can improve upon this paradigm by combining greater accuracy for high-level relative conformer energies with lower computational cost and more favorable scaling with molecular size. IR benchmarks against experimental gas-phase spectra additionally show performance comparable to quantum-chemical methods, with chemically interpretable mode assignments. Applications to alanine peptides, proline/β-cyclodextrin inclusion complexes, alanine dipeptide, and 2-methoxycyclohexanone further demonstrate that MARS captures conformational reorganization, noncovalent binding motifs, solvent-induced stability shifts, and spectroscopic signatures. These results establish MARS as a transferable and accessible platform for ensemble-based molecular modeling with machine-learned potentials. MARS is freely available at github.com/tcpunilu/mars.

  • Open Access

    Explainable ML force-field for evaluating protein–ligand binding energy using SO3LR

    by Hamza Agha, Sergio SuárEz-Dou, Adil Kabylda, Alexandre Tkatchenko, Andrea Volkamer

    Accurate estimation of protein–ligand binding affinities remains central to structure-based drug design. Classical force fields neglect important quantum-mechanical effects, while quantum mechanical and semi-empirical methods remain impractical for large biomolecular systems. Machine-learned force fields (MLFFs) offer an alternative, but their performance may be inflated by training on protein–ligand complexes. Here, we introduce SO3LR-SF, a physics-based scoring function built on the pretrained SO3LR MLFF, combining an equivariant message-passing network for semi-local interactions with physically motivated terms for short-range repulsion, electrostatics, and long-range dispersion. Built on an MLFF that was not trained on protein–ligand complexes, SO3LR-SF provides a stringent test of transferability. On PLA15 (15 active-site models), SO3LR-SF attains a relative interaction error of 5.8% against DLPNO-CCSD(T) references, the lowest of all tested methods. On the FEP benchmark (8 targets, 264 ligands), SO3LR-SF achieves an average Spearman correlation of 0.51, on par with the best semi-empirical quantum-mechanical method tested (0.47, GFN-FF) and with MMGBSA (0.44), and surpassing Glide (0.27). On the Wang dataset (8 targets, 199 ligands), it reaches 0.51, approaching the 0.60 average of molecular dynamics-based free-energy methods at a fraction of their cost. Each complex is scored in 1–10 seconds on 12 CPU cores, an 80- to 300-fold speedup over comparable MLFFs. An 8 Å trimming protocol reduces runtime threefold without loss of accuracy, and a restrained geometry optimization recovers ranking accuracy for targets with strained input structures. A multi-level explainability framework adds energy decomposition, per-atom energy contributions reported either for ligand or as 2D protein–ligand interaction maps, and 3D binding site visualization. We identified two pocket descriptors that flag applicability before scoring, e.g. polar solvent accessible surface area (SASA) ratio correlating strongly (r = 0.81) with performance. Together, SO3LR-SF presents a fast, interpretable, and competitive scoring function for drug discovery that requires no domain-specific training data.

  • Open Access

    MARS: Machine-Learned Force Field Framework for Automated Conformational Sampling, Vibrational Spectroscopy, and Microsolvation

    by Sergio Suárez-Dou, Miguel Gallegos, Alexandre Tkatchenko

    We present MARS, an open-source Python package for automated conformational ensemble generation, infrared spectrum prediction, and explicit microsolvation of molecular systems. Built on JAX for GPU acceleration and automatic differentiation, MARS is carried out with machine-learned force-fields (MLFFs), including SO3LR and MACE; with a four-phase conformational sampling workflow combining metadynamics, multi-temperature molecular dynamics, and genetic algorithms. The package provides automated vibrational mode classification using Wilson B-matrix projections and functional group identification via graph-based pattern matching. We benchmark MARS on chemically diverse systems including alkanes, halogenated aromatics, amino acids, peptides, drug-like molecules, phosphamides, host--guest complexes, and microsolvated biomolecular models. Semiempirical quantum-chemical methods have until now provided the practical state of the art for automated conformational sampling. Our benchmarks show that modern MLFFs can improve upon this paradigm by combining greater accuracy for high-level relative conformer energies with lower computational cost and more favorable scaling with molecular size. IR benchmarks against experimental gas-phase spectra additionally show performance comparable to quantum-chemical methods, with chemically interpretable mode assignments. Applications to alanine peptides, proline/β-cyclodextrin inclusion complexes, alanine dipeptide, and 2-methoxycyclohexanone further demonstrate that MARS captures conformational reorganization, noncovalent binding motifs, solvent-induced stability shifts, and spectroscopic signatures. These results establish MARS as a transferable and accessible platform for ensemble-based molecular modeling with machine-learned potentials. MARS is freely available at github.com/tcpunilu/mars .

  • Open Access

    QCell: Comprehensive Quantum-Mechanical Dataset Spanning Diverse Biomolecular Fragments

    by Adil Kabylda, Sergio Suárez-Dou, Nils Davoine, Florian N. Brünig, Alexandre Tkatchenko

    Recent advances in machine learning force fields (MLFFs) are revolutionizing molecular simulations by bridging the gap between quantum-mechanical (QM) accuracy and the computational efficiency of mechanistic potentials. However, the development of reliable MLFFs for biomolecular systems remains constrained by the scarcity of high-quality, chemically diverse QM datasets that span all of the major classes of biomolecules expressed in living cells. Crucially, such a comprehensive dataset must be computed using non-empirical or minimally empirical approximations to solving the Schrödinger equation. To address these limitations, we introduce the QCell dataset – a curated collection of 525k new QM calculations for biomolecular fragments encompassing carbohydrates, nucleic acids, lipids, dimers, and ion clusters. QCell complements existing datasets, bringing the total number of available data points to 41 million molecular systems, all calculated using hybrid density functional theory with nonlocal many-body dispersion interactions, as captured by the PBE0+MBD(-NL) level of quantum mechanics. The QCell dataset therefore provides a valuable resource for training next-generation MLFFs capable of modeling the intricate interactions that govern biomolecular dynamics beyond small molecules and proteins.


  • Top Cited Papers
    Open Access

    Molecular Simulations with a Pretrained Neural Network and Universal Pairwise Force Fields

    by Adil Kabylda, J. Thorben Frank, Sergio Suárez-Dou, Almaz Khabibrakhmanov, Leonardo Medrano Sandonas, Oliver T. Unke, Stefan Chmiela, Klaus-Robert Müller, Alexandre Tkatchenko

    Machine Learning Force Fields (MLFFs) promise to enable general molecular simulations that can simultaneously achieve efficiency, accuracy, transferability, and scalability for diverse molecules, materials, and hybrid interfaces. A key step toward this goal has been made with the GEMS approach to biomolecular dynamics [Unke et al., Sci. Adv. 2024, 10, eadn4397]. This work introduces the SO3LR method that integrates the fast and stable SO3krates neural network for semilocal interactions with universal pairwise force fields designed for short-range repulsion, long-range electrostatics, and dispersion interactions. SO3LR is trained on a diverse set of 4 million neutral and charged molecular complexes computed at the PBE0+MBD level of quantum mechanics, ensuring broad coverage of covalent and noncovalent interactions. Our approach is characterized by computational and data efficiency, scalability to 200 thousand atoms on a single GPU, and reasonable to high accuracy across the chemical space of organic (bio)molecules. SO3LR is applied to study units of four major biomolecule types, polypeptide folding, and nanosecond dynamics of larger systems such as a protein, a glycoprotein, and a lipid bilayer, all in explicit solvent. Finally, we discuss future challenges toward truly general molecular simulations by combining MLFFs with traditional atomistic models.


  • Open Access

    Molecular Simulations with a Pretrained Neural Network and Universal Pairwise Force Fields

    by Adil Kabylda, J. Thorben Frank, Sergio Suarez Dou, Almaz Khabibrakhmanov, Leonardo Medrano Sandonas, Oliver T. Unke, Stefan Chmiela, Klaus-Robert Müller, Alexandre Tkatchenko

    Machine Learning Force Fields (MLFFs) promise to enable general molecular simulations that can simultaneously achieve efficiency, accuracy, transferability, and scalability for diverse molecules, materials, and hybrid interfaces. A key step toward this goal has been made with the GEMS approach to biomolecular dynamics [Sci. Adv. 10, eadn4397 (2024)]. This work introduces the SO3LR method that integrates the fast and stable SO3krates neural network for semi-local interactions with universal pairwise force fields designed for short-range repulsion, long-range electrostatics, and dispersion interactions. SO3LR is trained on a diverse set of 4 million neutral and charged molecular complexes computed at the PBE0+MBD level of quantum mechanics, ensuring broad coverage of covalent and non-covalent interactions. Our approach is characterized by computational and data efficiency, scalability to 200 thousand atoms on a single GPU, and reasonable to high accuracy across the chemical space of organic (bio)molecules. SO3LR is applied to study units of four major biomolecule types, polypeptide folding, and nanosecond dynamics of larger systems such as a protein, a glycoprotein, and a lipid bilayer, all in explicit solvent. Finally, we discuss future challenges toward truly general molecular simulations by combining MLFFs with traditional atomistic models.


  • Top Cited Papers
    Open Access

    Large scale investigation of GPCR molecular dynamics data uncovers allosteric sites and lateral gateways

    by David Aranda-García, Tomasz Maciej Stepniewski, Mariona Torrens-Fontanals, Adrian García-Recio, Marta Lopez-Balastegui, Brian Medel-Lacruz, Adrián Morales-Pastor, Alejandro Peralta-García, Miguel Dieguez-Eceolaza, David Sotillo-Nuñez, et al.

    G protein-coupled receptors (GPCRs) constitute a functionally diverse protein family and are targets for a broad spectrum of pharmaceuticals. Technological progress in X-ray crystallography and cryogenic electron microscopy has enabled extensive, high-resolution structural characterisation of GPCRs in different conformational states. However, as highly dynamic events underlie GPCR signalling, a complete understanding of GPCR functionality requires insights into their conformational dynamics. Here, we present a large dataset of molecular dynamics simulations covering 60% of currently available GPCR structures. Our analysis reveals extensive local “breathing” motions of the receptor on a nano- to microsecond timescale and provides access to numerous previously unexplored receptor conformational states. Furthermore, we reveal that receptor flexibility impacts the shape of allosteric drug binding sites, which frequently adopt partially or completely closed states in the absence of a molecular modulator. We demonstrate that exploring membrane lipid dynamics and their interaction with GPCRs is an efficient approach to expose such hidden allosteric sites and even lateral ligand entrance gateways. The obtained insights and generated dataset on conformations, allosteric sites and lateral entrance gates in GPCRs allows us to better understand the functionality of these receptors and opens new therapeutic avenues for drug-targeting strategies.