A5101829262- Machine Learning in Materials Science
- Protein Structure and Dynamics
- Computational Drug Discovery Methods
- Electron and X-Ray Spectroscopy Techniques
- Ionic liquids properties and applications
- DNA and Nucleic Acid Chemistry
- Advanced biosensing and bioanalysis techniques
- Molecular Junctions and Nanostructures
- Spectroscopy and Quantum Chemical Studies
- Various Chemistry Research Topics
- Enzyme Structure and Function
- Machine Learning in Bioinformatics
- Nanopore and Nanochannel Transport Studies
- Block Copolymer Self-Assembly
- Electrochemical Analysis and Applications
- RNA Interference and Gene Delivery
- Tensor decomposition and applications
- Advanced Optimization Algorithms Research
- Mathematical Biology Tumor Growth
- Electronic and Structural Properties of Oxides
- DNA Repair Mechanisms
- Phase Equilibria and Thermodynamics
- Fuel Cells and Related Materials
- Topic Modeling
- Advanced Chemical Physics Studies
Massachusetts Institute of Technology
2019-2023
Intarcia Therapeutics (United States)
2022-2023
Moscow Institute of Thermal Technology
2022
Harvard University
2020
Wesleyan University
2016-2017
Abstract Three billion years of evolution has produced a tremendous diversity protein molecules 1 , but the full potential proteins is likely to be much greater. Accessing this been challenging for both computation and experiments because space possible larger than those have functions. Here we introduce Chroma, generative model complexes that can directly sample novel structures sequences, conditioned steer process towards desired properties To enable this, diffusion respects conformational...
Abstract Molecular dynamics simulations provide theoretical insight into the microscopic behavior of condensed-phase materials and, as a predictive tool, enable computational design new compounds. However, because large spatial and temporal scales thermodynamic kinetic phenomena in materials, atomistic are often computationally infeasible. Coarse-graining methods allow larger systems to be simulated by reducing their dimensionality, propagating longer timesteps, averaging out fast motions....
Molecular dynamics (MD) simulation techniques are widely used for various natural science applications. Increasingly, machine learning (ML) force field (FF) models begin to replace ab-initio simulations by predicting forces directly from atomic structures. Despite significant progress in this area, such primarily benchmarked their force/energy prediction errors, even though the practical use case would be produce realistic MD trajectories. We aim fill gap introducing a novel benchmark suite...
Abstract Three billion years of evolution have produced a tremendous diversity protein molecules, and yet the full potential this molecular class is likely far greater. Accessing has been challenging for computation experiments because space possible molecules much larger than those to host function. Here we introduce Chroma, generative model proteins complexes that can directly sample novel structures sequences be conditioned steer process towards desired properties functions. To enable...
Learning pair interactions from experimental or simulation data is of great interest for molecular simulations. We propose a general stochastic method learning using differentiable simulations (DiffSim). DiffSim defines loss function based on structural observables, such as the radial distribution function, through dynamics (MD) The interaction potentials are then learned directly by gradient descent, backpropagation to calculate metric with respect potential MD simulation. This...
Solvate ionic liquids (SIL) have promising applications as electrolyte materials and machine learning can help accelerate the virtual screening of candidate molecules for SIL.
Molecular dynamics simulations use statistical mechanics at the atomistic scale to enable both elucidation of fundamental mechanisms and engineering matter for desired tasks. The behavior molecular systems microscale is typically simulated with differential equations parameterized by a Hamiltonian, or energy function. Hamiltonian describes state system its interactions environment. In order derive predictive microscopic models, one wishes infer that agrees observed macroscopic quantities....
Computer simulations can provide mechanistic insight into ionic liquids (ILs) and predict the properties of experimentally unrealized ion combinations. However, ILs suffer from a particularly large disparity in time scales atomistic ensemble motion. Coarse-grained models are therefore used place costly simulations, allowing simulation longer larger systems. Nevertheless, constructing many-body potential mean force that defines structure dynamics coarse-grained system be complicated...
Abstract Holliday junctions play a central role in genetic recombination, DNA repair and other cellular processes. We combine simulations experiments to evaluate the ability of 3SPN.2 model, coarse-grained representation designed mimic B-DNA, predict properties junctions. The model reproduces many experimentally determined aspects junction structure stability, including temperature dependence melting on salt concentration, bias between open stacked conformations, relative populations...
Coarse-graining (CG) of molecular simulations simplifies the particle representation by grouping selected atoms into pseudo-beads and drastically accelerates simulation. However, such CG procedure induces information losses, which makes accurate backmapping, i.e., restoring fine-grained (FG) coordinates from coordinates, a long-standing challenge. Inspired recent progress in generative models equivariant networks, we propose novel model that rigorously embeds vital probabilistic nature...
Solvate Ionic Liquids (SIL) have promising applications as electrolyte materials. Despite the broad design space of oligoether ligands, most reported SILs are based on simple tri- and tetraglyme. Here, we describe a computational search for complex ethers that can better stabilize SILs. Through active learning, neural network interatomic potential is trained from density functional theory data. The learned fulfills two key requirements: transferability across composition space, high speed...
Solvate Ionic Liquids (SIL) have promising applications as electrolyte materials. Despite the broad design space of oligoether ligands, most reported SILs are based on simple tri- and tetraglyme. Here, we describe a computational search for complex ethers that can better stabilize SILs. Through active learning, neural network interatomic potential is trained from density functional theory data. The learned fulfills two key requirements: transferability across composition space, high speed...
Modeling dynamical effects in chemical reactions, such as post-transition state bifurcation, requires ab initio molecular dynamics simulations due to the breakdown of simpler static models like transition theory. However, these tend be restricted lower-accuracy electronic structure methods and scarce sampling because their high computational cost. Here, we report use statistical learning accelerate reactive by combining high-throughput calculations, graph-convolution interatomic potentials...
<div>Modeling dynamical effects in chemical reactions, such as post-transition state bifurcation, requires <i>ab initio</i> molecular dynamics simulations due to the breakdown of simpler static models like transition theory. However, these tend be restricted lower-accuracy electronic structure methods and scarce sampling because their high computational cost. Here, we report use statistical learning accelerate reactive by combining high-throughput ab initio calculations,...
Learning pair interactions from experimental or simulation data is of great interest for molecular simulations. We propose a general stochastic method learning using differentiable simulations (DiffSim). DiffSim defines loss function based on structural observables, such as the radial distribution function, through dynamics (MD) The interaction potentials are then learned directly by gradient descent, backpropagation to calculate metric with respect potential MD simulation. This...
Predicting molecular conformations (or 3D structures) from graphs is a fundamental problem in many applications. Most existing approaches are usually divided into two steps by first predicting the distances between atoms and then generating structure through optimizing distance geometry problem. However, predicted with such two-stage may not be able to consistently preserve of local atomic neighborhoods, making generated structures unsatisfying. In this paper, we propose an end-to-end...
Modeling dynamical effects in chemical reactions, such as post-transition state bifurcation, requires ab initio molecular dynamics simulations due to the breakdown of simpler static models like transition theory. However, these tend be restricted lower-accuracy electronic structure methods and scarce sampling because their high computational cost. Here, we report use statistical learning accelerate reactive by combining high-throughput calculations, graph-convolution interatomic potentials...