A5027026045- Protein Structure and Dynamics
- Computational Drug Discovery Methods
- Machine Learning in Materials Science
- Enzyme Structure and Function
- Receptor Mechanisms and Signaling
- Colorectal Cancer Treatments and Studies
- Lung Cancer Treatments and Mutations
- Spectroscopy and Quantum Chemical Studies
- Mass Spectrometry Techniques and Applications
- Radiomics and Machine Learning in Medical Imaging
- Bioinformatics and Genomic Networks
- Lipid Membrane Structure and Behavior
- Reinforcement Learning in Robotics
- Nanopore and Nanochannel Transport Studies
- Chemical Synthesis and Analysis
- Neuropeptides and Animal Physiology
- Material Dynamics and Properties
- Polyomavirus and related diseases
- Advanced Bandit Algorithms Research
- Block Copolymer Self-Assembly
- RNA and protein synthesis mechanisms
- Gene Regulatory Network Analysis
- Genetics, Bioinformatics, and Biomedical Research
- RNA Research and Splicing
- Monoclonal and Polyclonal Antibodies Research
Barcelona Biomedical Research Park
2016-2025
Universitat Pompeu Fabra
2016-2025
Institució Catalana de Recerca i Estudis Avançats
2016-2025
Acellera (Spain)
2017-2025
Universität Ulm
2024
Helmholtz Zentrum München
2024
Institute of Molecular Biology
2024
Center for Integrated Protein Science Munich
2024
Technical University of Munich
2024
Stanford University
2020
The high arithmetic performance and intrinsic parallelism of recent graphical processing units (GPUs) can offer a technological edge for molecular dynamics simulations. ACEMD is production-class biomolecular (MD) engine supporting CHARMM AMBER force fields. Designed specifically GPUs it able to achieve supercomputing scale 40 ns/day all-atom protein systems with over 23 000 atoms. We provide validation evaluation the code run microsecond-long trajectory an system in explicit TIP3P water on...
Accurately predicting protein–ligand binding affinities is an important problem in computational chemistry since it can substantially accelerate drug discovery for virtual screening and lead optimization. We propose here a fast machine-learning approach using state-of-the-art 3D-convolutional neural networks compare this to other scoring methods several diverse data sets. The results the standard PDBbind (v.2016) core test-set are with Pearson's correlation coefficient of 0.82 RMSE 1.27 pK...
A goal in the kinetic characterization of a macromolecular system is description its slow relaxation processes via (i) identification structural changes involved these and (ii) estimation rates or timescales at which occur. Most approaches to this task, including Markov models, master-equation network start by discretizing high-dimensional state space then characterize terms eigenvectors eigenvalues discrete transition matrix. The practical success such an approach depends very much on...
The understanding of protein–ligand binding is critical importance for biomedical research, yet the process itself has been very difficult to study because its intrinsically dynamic character. Here, we have able quantitatively reconstruct complete enzyme-inhibitor complex trypsin-benzamidine by performing 495 molecular dynamics simulations free ligand 100 ns each, 187 which produced events with an rmsd less than 2 Å compared crystal structure. paths obtained are capture kinetic pathway...
Abstract Motivation An important step in structure-based drug design consists the prediction of druggable binding sites. Several algorithms for detecting cavities, those likely to bind a small compound, have been developed over years by clever exploitation geometric, chemical and evolutionary features protein. Results Here we present novel knowledge-based approach that uses state-of-the-art convolutional neural networks, where algorithm is learned examples. In total, 7622 proteins from scPDB...
Atomistic or ab initio molecular dynamics simulations are widely used to predict thermodynamics and kinetics relate them structure. A common approach go beyond the time- length-scales accessible with such computationally expensive is definition of coarse-grained models. Existing coarse-graining approaches define an effective interaction potential match defined properties high-resolution models experimental data. In this paper, we reformulate as a supervised machine learning problem. We use...
Recent advances in molecular simulations have allowed scientists to investigate slower biological processes than ever before. Together with these came an explosion of data that has transformed a traditionally computing-bound into data-bound problem. Here, we present HTMD, programmable, extensible platform written Python aims solve the generation and analysis problem as well increase reproducibility by providing complete workspace for simulation-based discovery. So far, HTMD includes system...
The smooth particle mesh Ewald summation method is widely used to efficiently compute long-range electrostatic force terms in molecular dynamics simulations, and there has been considerable work developing optimized implementations for a variety of parallel computer architectures. We describe an implementation Nvidia graphical processing units (GPUs) which are general purpose computing devices with high degree intrinsic parallelism arithmetic performance. find that, typical biomolecular...
Protein preparation is a critical step in molecular simulations that consists of refining Data Bank (PDB) structure by assigning titration states and optimizing the hydrogen-bonding network. In this application note, we describe ProteinPrepare, web designed to interactively support protein structures. Users can upload PDB file, choose solvent pH value, inspect resulting protonated residues network within 3D interface. Protonation are suggested automatically but be manually changed using...
Abstract Purpose: Patients with colorectal cancer who respond to the anti-EGFR antibody cetuximab often develop resistance within several months of initiating therapy. To design new lines treatment, molecular landscape resistant tumors must be ascertained. We investigated role mutations in EGFR signaling axis on acquisition patients and cellular models. Experimental Design: Tissue samples were obtained from 37 became refractory cetuximab. Colorectal cells sensitive treated until derivatives...
In this work, we propose a machine learning approach to generate novel molecules starting from seed compound, its three-dimensional (3D) shape, and pharmacophoric features. The pipeline draws inspiration generative models used in image analysis represents first example of the de novo design lead-like guided by shape-based A variational autoencoder is perturb 3D representation followed system convolutional recurrent neural networks that sequence SMILES tokens. scaffolds functional groups can...
Molecular dynamics simulations provide a mechanistic description of molecules by relying on empirical potentials. The quality and transferability such potentials can be improved leveraging data-driven models derived with machine learning approaches. Here, we present TorchMD, framework for molecular mixed classical All force computations including bond, angle, dihedral, Lennard-Jones, Coulomb interactions are expressed as PyTorch arrays operations. Moreover, TorchMD enables simulating neural...
Coarse graining enables the investigation of molecular dynamics for larger systems and at longer timescales than is possible atomic resolution. However, a coarse model must be formulated such that conclusions we draw from it are consistent with would finer level detail. It has been proven force matching scheme defines thermodynamically coarse-grained an atomistic system in variational limit. Wang et al. [ACS Cent. Sci. 5, 755 (2019)] demonstrated existence limit use supervised machine...
Machine learning plays an important and growing role in molecular simulation. The newest version of the OpenMM dynamics toolkit introduces new features to support use machine potentials. Arbitrary PyTorch models can be added a simulation used compute forces energy. A higher-level interface allows users easily model their molecules interest with general purpose, pretrained potential functions. collection optimized CUDA kernels custom operations greatly improves speed simulations. We...
Abstract Machine learning potentials are an important tool for molecular simulation, but their development is held back by a shortage of high quality datasets to train them on. We describe the SPICE dataset, new quantum chemistry dataset training relevant simulating drug-like small molecules interacting with proteins. It contains over 1.1 million conformations diverse set molecules, dimers, dipeptides, and solvated amino acids. includes 15 elements, charged uncharged wide range covalent...
Machine learning potentials have emerged as a means to enhance the accuracy of biomolecular simulations. However, their application is constrained by significant computational cost arising from vast number parameters compared with traditional molecular mechanics. To tackle this issue, we introduce an optimized implementation hybrid method (NNP/MM), which combines neural network potential (NNP) and mechanics (MM). This approach models portion system, such small molecule, using NNP while...
Abstract A generalized understanding of protein dynamics is an unsolved scientific problem, the solution which critical to interpretation structure-function relationships that govern essential biological processes. Here, we approach this problem by constructing coarse-grained molecular potentials based on artificial neural networks and grounded in statistical mechanics. For training, build a unique dataset unbiased all-atom simulations approximately 9 ms for twelve different proteins with...
This letter gives results on improving protein-ligand binding affinity predictions based molecular dynamics simulations using machine learning potentials with a hybrid neural network potential and mechanics methodology (NNP/MM). We compute relative free energies (RBFE) the Alchemical Transfer Method (ATM) validate its performance against established benchmarks find significant enhancements compared to conventional MM force fields like GAFF2.
Achieving a balance between computational speed, prediction accuracy, and universal applicability in molecular simulations has been persistent challenge. This paper presents substantial advancements TorchMD-Net software, pivotal step forward the shift from conventional force fields to neural network-based potentials. The evolution of into more comprehensive versatile framework is highlighted, incorporating cutting-edge architectures such as TensorNet. transformation achieved through modular...
Although molecular dynamics simulation methods are useful in the modeling of macromolecular systems, they remain computationally expensive, with production work requiring costly high-performance computing (HPC) resources. We review recent innovations accelerating on graphics processing units (GPUs), and we describe GPUGRID, a volunteer project that uses GPU resources nondedicated desktop workstation computers. In particular, demonstrate capability simulating thousands all-atom trajectories...
High-throughput molecular dynamics (MD) simulations are a computational method consisting of using multiple short trajectories, instead few long ones, to cover slow biological time scales. Compared trajectories this offers the possibility start in successive batches, building knowledgeable model available data inform subsequent new iteratively. Here, we demonstrate an automatic, iterative, on-the-fly for learning and sampling context ligand binding case trypsin-benzamidine binding. The uses...
Chemical space is impractically large, and conventional structure-based virtual screening techniques cannot be used to simply search through the entire discover effective bioactive molecules. To address this shortcoming, we propose a generative adversarial network generate, rather than search, diverse three-dimensional ligand shapes complementary pocket. Furthermore, show that generated molecule can decoded using shape-captioning into sequence of SMILES enabling directly de novo drug design....