Redirecting to original paper in 30 seconds...

Click below to go immediately or wait for automatic redirect

arxiv_ml 95% Match Research Paper Computational chemists,Biophysicists,Drug discovery scientists,Machine learning researchers 1 week ago

JAMUN: Bridging Smoothed Molecular Dynamics and Score-Based Learning for Conformational Ensembles

graph-neural-networks › molecular-modeling
📄 Abstract

Abstract: Conformational ensembles of protein structures are immensely important both for understanding protein function and drug discovery in novel modalities such as cryptic pockets. Current techniques for sampling ensembles such as molecular dynamics (MD) are computationally inefficient, while many recent machine learning methods do not transfer to systems outside their training data. We propose JAMUN which performs MD in a smoothed, noised space of all-atom 3D conformations of molecules by utilizing the framework of walk-jump sampling. JAMUN enables ensemble generation for small peptides at rates of an order of magnitude faster than traditional molecular dynamics. The physical priors in JAMUN enables transferability to systems outside of its training data, even to peptides that are longer than those originally trained on. Our model, code and weights are available at https://github.com/prescient-design/jamun.
Authors (6)
Ameya Daigavane
Bodhi P. Vani
Darcy Davidson
Saeed Saremi
Joshua Rackers
Joseph Kleinhenz
Submitted
October 18, 2024
arXiv Category
physics.bio-ph
arXiv PDF Code

Key Contributions

JAMUN bridges smoothed molecular dynamics and score-based learning to efficiently generate conformational ensembles of protein structures. This approach significantly accelerates ensemble generation compared to traditional molecular dynamics and exhibits improved transferability to unseen systems, which is crucial for drug discovery and understanding protein function.

Business Value

Accelerates the drug discovery process by enabling faster and more accurate simulation of protein conformations, potentially leading to the identification of novel drug targets and therapies.

Paper Metadata

Innovation Type

Algorithmic

Deployment Feasibility

Requires significant computational resources for simulations, but the speedup offers practical advantages. Availability of code and weights aids adoption.

Limitations Addressed

Computational inefficiency of traditional molecular dynamics for ensemble sampling and lack of transferability in existing machine learning methods to systems outside their training data.

Performance Gains

Order of magnitude faster ensemble generation than traditional molecular dynamics.

View Code on GitHub

Technical Tags

molecular dynamicsscore-based learningconformational ensembleswalk-jump samplingsmoothed molecular dynamicsall-atom 3D conformationsprotein structuresdrug discoverycryptic pocketspeptides

Research Topics

Computational ChemistryMachine Learning for ScienceMolecular SimulationDrug DiscoveryProtein Folding

Methods & Architectures

Walk-jump samplingSmoothed molecular dynamicsScore-based learning Score-based generative models

Applications & Tasks

Drug Discovery Biotechnology Computational Biology Sampling molecular conformationsAccelerating molecular dynamics simulationsPredicting protein functionIdentifying drug targets Generating conformational ensemblesDrug discoveryUnderstanding protein function

Related Fields

Computational ChemistryBiophysicsMachine LearningPharmacology

Keywords

JAMUNmolecular dynamicsscore-based learningconformational ensemblesprotein structuresdrug discoverycryptic pocketswalk-jump samplingsmoothed spacetransferabilitycomputational efficiencybiotechnologypharmacology

Academic Context

#Computational Chemistry#Machine Learning for Science#Molecular Simulation#Drug Discovery#Protein Folding

Companies & Organizations

Companies Mentioned
Prescient Design
Startup Context
Prescient Design

Commercial Potential

Potential Products
Accelerated molecular simulation softwareDrug target identification platforms
Target Industries
PharmaceuticalsBiotechnologyChemicals
Use Case Examples
Simulating protein folding dynamicsIdentifying binding sites for drug candidatesUnderstanding protein-ligand interactions
Competitive Edge
Offers a significant speedup and improved transferability over traditional MD and other ML methods for conformational ensemble generation.
Market Opportunity
Large market for drug discovery and molecular simulation tools.
Revenue Models
Software licensingcloud-based simulation services.

Resource Requirements

Compute Needs
High-performance computing for molecular dynamics simulations.
Data Requirements
Molecular structures, potentially experimental data for validation.
Deployment Constraints
Requires specialized expertise in molecular simulation and ML.
Scalability
Enables faster generation for small peptides; scalability to larger proteins needs further investigation.

Production Readiness

Maturity Level

Research

Time to Market

1-3 years for integration into drug discovery pipelines.

Licensing

Code and weights are available, suggesting open-source or permissive licensing.

View Full Paper Back to Papers