Differentiable breeding: Automatic differentiation enables efficient gradient-based optimization of breeding strategies DOI Creative Commons
Kosuke Hamazaki, Hiroyoshi Iwata, Koji Tsuda

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 3, 2024

Abstract Conventional breeding methods often require extensive time to develop new cultivars, hindering rapid adaptation global challenges. While genomic selection has accelerated breeding, there remains substantial room for improvement. Recent studies have explored complex decision-making in schemes using various optimization techniques such as black-box optimization. However, these are challenged by constraints simultaneously optimizing multiple parameters necessary achieving more efficient and flexible To address limitations, this study implemented automatic differentiation of PyTorch. By treating the entire scheme a differentiable computational graph, we enabled gradient calculations final genetic gains relative progeny allocation each mating pair. We first validated our approaches comparing with analytical results simple gamete generation test case. Next, used perform gradient-based strategies, aiming maximize schemes. The strategy was then compared black-box-based optimized non-optimized strategies. Our framework successfully reduced number function evaluations needed approach outperformed terms gains. This demonstrates that effectively harnessed information via differentiation. Integrating into is expected enhance flexibility lay groundwork future methods. Author summary Plant plays crucial role addressing challenges like population growth climate change developing adaptable crop varieties. conventional several years produce making it difficult keep pace advancements techniques, selection, significantly enhanced accuracy speed. Despite improvements, real programs. explores application optimize specifically focusing on progenies allocated Automatic enables calculation derivatives functions, potentially accelerating process based PyTorch, graph. integrating optimization, enable exploration optimal solutions while greater parameters. novel method potential efficiency ultimately contributing development productive varieties food

Language: Английский

Application of Modern Artificial Intelligence Techniques in the Development of Organic Molecular Force Fields DOI
Junmin Chen, Qian Gao,

Miaofei Huang

et al.

Physical Chemistry Chemical Physics, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

The molecular force field (FF) determines the accuracy of dynamics (MD) and is one major bottlenecks that limits application MD in design. Recently, artificial intelligence (AI) techniques, such as machine-learning potentials (MLPs), have been rapidly reshaping landscape MD. Meanwhile, organic systems feature unique characteristics, require more careful treatment both model construction, optimization, validation. While an accurate generic still missing, significant progress has made with facilitation AI, warranting a promising future. In this review, we provide overview various types AI techniques used FF development discuss advantages weaknesses these methodologies. We show how methods unprecedented capabilities many tasks potential fitting, atom typification, automatic optimization. it also worth noting efforts are needed to improve transferability model, develop comprehensive database, establish standardized validation procedures. With discussions, hope inspire solve existing problems, eventually leading birth next-generation FFs.

Language: Английский

Citations

1
Exploring parameter dependence of atomic minima with implicit differentiation DOI Creative Commons
Ivan Maliyov, Petr Grigorev,

Thomas D. Swinburne

et al.

npj Computational Materials, Journal Year: 2025, Volume and Issue: 11(1)

Published: Jan. 27, 2025

Language: Английский

Citations

1
Scaling Graph Neural Networks to Large Proteins DOI
Justin Airas, Bin Zhang

Journal of Chemical Theory and Computation, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 6, 2025

Graph neural network (GNN) architectures have emerged as promising force field models, exhibiting high accuracy in predicting complex energies and forces based on atomic identities Cartesian coordinates. To expand the applicability of GNNs, machine learning fields more broadly, optimizing their computational efficiency is critical, especially for large biomolecular systems classical molecular dynamics simulations. In this study, we address key challenges existing GNN benchmarks by introducing a dataset, DISPEF, which comprises large, biologically relevant proteins. DISPEF includes 207,454 proteins with sizes up to 12,499 atoms features diverse chemical environments, spanning folded disordered regions. The implicit solvation free energies, used training targets, represent particularly challenging case due many-body nature, providing stringent test evaluating expressiveness models. We benchmark performance seven GNNs emphasizing importance directly accounting long-range interactions enhance model transferability. Additionally, present novel multiscale architecture, termed Schake, delivers transferable computationally efficient energy predictions Our findings offer valuable insights tools advancing protein modeling applications.

Language: Английский

Citations

1
The Open Force Field Initiative: Open Software and Open Science for Molecular Modeling DOI Creative Commons
Lily Wang, Pavan Kumar Behara, Matthew W. Thompson

et al.

The Journal of Physical Chemistry B, Journal Year: 2024, Volume and Issue: unknown

Published: July 11, 2024

Force fields are a key component of physics-based molecular modeling, describing the energies and forces in system as function positions atoms molecules involved. Here, we provide review scientific status report on work Open Field (OpenFF) Initiative, which focuses science, infrastructure data required to build next generation biomolecular force fields. We introduce OpenFF Initiative related Consortium, describe its approach field development software, discuss accomplishments date well future plans. releases both software under open permissive licensing agreements enable rapid application, validation, extension, modification tools. lessons learned this new development. also highlight ways that other researchers can get involved, some recent successes outside taking advantage tools data.

Language: Английский

Citations

7
Metal–Organic Frameworks through the Lens of Artificial Intelligence: A Comprehensive Review DOI

Kevizali Neikha,

Амрит Пузари

Langmuir, Journal Year: 2024, Volume and Issue: 40(42), P. 21957 - 21975

Published: Oct. 9, 2024

Metal–organic frameworks (MOFs) are a class of hybrid porous materials that have gained prominence as noteworthy material with varied applications. Currently, MOFs in extensive use, particularly the realms energy and catalysis. The synthesis these poses considerable challenges, their computational analysis is notably intricate due to complex structure versatile applications field science. Density functional theory (DFT) has helped researchers understanding reactions mechanisms, but it costly time-consuming requires bigger systems perform calculations. Machine learning (ML) techniques were adopted order overcome problems by implementing ML data sets for synthesis, structure, property predictions MOFs. These fast, efficient, accurate do not require heavy computing. In this review, we discuss models used MOF incorporation artificial intelligence (AI) predictions. advantage AI would accelerate research, synthesizing novel multiple properties oriented minimum information.

Language: Английский

Citations

5
Transient Subtraction: A Control Variate Method for Computing Transport Coefficients DOI
Pierre Monmarché, Renato Spacek, Gabriel Stoltz

et al.

Journal of Statistical Physics, Journal Year: 2025, Volume and Issue: 192(4)

Published: April 5, 2025

Language: Английский

Citations

0
Fine-tuning molecular mechanics force fields to experimental free energy measurements DOI Creative Commons
Dominic A. Rufa, Josh Fass, John D. Chodera

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 8, 2025

Alchemical free energy methods using molecular mechanics (MM) force fields are essential tools for predicting thermodynamic properties of small molecules, especially via calculations that can estimate quantities relevant drug discovery such as affinities, selectivities, the impact target mutations, and ADMET properties. While traditional MM forcefields rely on hand-crafted, discrete atom types parameters, modern approaches based graph neural networks (GNNs) learn continuous embedding vectors represent chemical environments from which parameters be generated. Excitingly, GNN parameterization provide a fully end-to-end differentiable model offers possibility systematically improving these models experimental data. In this study, we treat pretrained field-here, espaloma-0.3.2-as foundation simulation fine-tune its charge limited hydration data, with goal assessing degree to improve prediction other related energies. We demonstrate highly efficient "one-shot fine-tuning" method an exponential (Zwanzig) reweighting estimator accuracy without need resimulate configurations. To achieve "one-shot" improvement, importance effective sample size (ESS) regularization strategies retain good overlap between initial fine-tuned fields. Moreover, show leveraging low-rank projections comparable improvements higher-dimensional in variety data-size regimes. Our results linearly-perturbative fine-tuning electrostatic data cost-effective strategy achieves state-of-the-art performance energies FreeSolv dataset.

Language: Английский

Citations

0
Molecular mechanism of Mad2 conformational conversion promoted by the Mad2‐interaction motif of Cdc20 DOI Creative Commons
C. Yu, Elyse S. Fischer, Joe G. Greener

et al.

Protein Science, Journal Year: 2025, Volume and Issue: 34(4)

Published: March 27, 2025

During mitosis, unattached kinetochores trigger the spindle assembly checkpoint by promoting of mitotic complex, a heterotetramer comprising Mad2, Cdc20, BubR1, and Bub3. Critical to this process is kinetochore-mediated catalysis an intrinsically slow conformational conversion Mad2 from open (O-Mad2) inactive state closed (C-Mad2) active bound Cdc20. These changes involve substantial remodeling N-terminal β1 strand C-terminal β7/β8 hairpin. In vitro, Mad2-interaction motif (MIM) Cdc20 (Cdc20MIM) triggers rapid O-Mad2 C-Mad2, effectively removing kinetic barrier for MCC assembly. How Cdc20MIM directly induces remains unclear. study, we demonstrate that Cdc20MIM-binding site inaccessible in O-Mad2. Time-resolved NMR molecular dynamics simulations show how involves sequential flexible structural elements O-Mad2, orchestrated Cdc20MIM. Conversion initiated hairpin transiently unfolding expose nascent site. Engagement promotes release strand. We propose initial allow binding transient intermediate thereby lowering conversion.

Language: Английский

Citations

0
Learning Force Field Parameters from Differentiable Particle-Field Molecular Dynamics DOI Creative Commons
Manuel Carrer, Henrique Musseli Cezar, Sigbjørn Løland Bore

et al.

Journal of Chemical Information and Modeling, Journal Year: 2024, Volume and Issue: 64(14), P. 5510 - 5520

Published: July 4, 2024

We develop ∂-HylleraasMD (∂-HyMD), a fully end-to-end differentiable molecular dynamics software based on the Hamiltonian hybrid particle-field formalism, and use it to establish protocol for automated optimization of force field parameters. ∂-HyMD is templated recently released HylleraaasMD software, while using JAX autodiff framework as main engine dynamics. exploits an embarrassingly parallel algorithm by spawning independent simulations, whose trajectories are simultaneously processed reverse mode automatic differentiation calculate gradient loss function, which in turn used iterative force-field show that organization facilitates convergence minimization procedure, avoiding known memory numerical stability issues approaches. showcase effectiveness our implementation producing library parameters standard phospholipids, with either zwitterionic or anionic heads saturated unsaturated tails. Compared all-atom reference, obtained yields better density profiles than derived from previously utilized gradient-free procedures. Moreover, models can predict good accuracy properties not included learning objective, such lateral pressure profiles, transferable other systems, including triglycerides.

Language: Английский

Citations

3
Application of Nosé–Hoover dynamics for coarse-graining molecular systems: An evaluation of reproducibility in Lennard-Jones systems DOI
Toru Yamada, Yohei MORINISHI

The Journal of Chemical Physics, Journal Year: 2025, Volume and Issue: 162(13)

Published: April 3, 2025

This study proposes an application of Nosé-Hoover (NH) dynamics as a coarse-graining (CG) method for molecular simulations, offering alternative to traditional Langevin-based approaches. The NH dynamics, known its deterministic temperature control without stochastic forces, is adapted here model monoatomic Lennard-Jones system at different coarse-grained levels. CG particle's equation motion derived from atomic-level linking thermostat terms with properties obtained (MD) simulations. Key parameters, including the coefficient and thermal inertia, are calibrated using MD data assess their impact on dynamic structural reproducibility model. calibration results suggest potential method. effectiveness proposed then evaluated through set show stable energy regulation promising accuracy in reproducing properties, particularly mass diffusion, opportunities further refinement representing momentum diffusion.

Language: Английский

Citations

0