CHARMM at 45: Enhancements in Accessibility, Functionality, and Speed

Wonmuk Hwang, Steven L. Austin, Arnaud Blondel, Eric D. Boittier, Stefan Boresch, Matthias Buck, Joshua Buckner, Amedeo Caflisch, Hao-Ting Chang, Xi Cheng, Yeol Kyo Choi, Jhih‐Wei Chu, Michael F. Crowley, Qiang Cui, A. Damjanović, Yuqing Deng, M. Devereux, Xinqiang Ding, Michael Feig, Jiali Gao, David R. Glowacki, James E. Gonzales, Mehdi Bagheri Hamaneh, Edward Harder, Ryan L. Hayes, Jing Huang, Yandong Huang, Phillip S. Hudson, Wonpil Im, Shahidul M. Islam, Wei Jiang, Michael R. Jones, Silvan Käser, Fiona L. Kearns, Nathan R. Kern, Jeffery B. Klauda, Themis Lazaridis, Jinhyuk Lee, Justin A. Lemkul, Xiaorong Liu, Yun Luo, Alexander D. MacKerell, Dan Thomas Major, Markus Meuwly, Kwangho Nam, Lennart Nilsson, Victor Ovchinnikov, Emanuele Paci, Soohyung Park, Richard W. Pastor, Amanda R. Pittman, Carol Beth Post, Samarjeet Prasad, Jingzhi Pu, Yifei Qi, Thenmalarchelvi Rathinavelan, Daniel R. Roe, Benoı̂t Roux, Christopher N. Rowley, Jana Shen, Andrew C. Simmonett, Alexander J. Sodt, Kai Töpfer, Meenu Upadhyay, Arjan van der Vaart, Luis Itza Vazquez-Salazar, Richard M. Venable, Luke Warrensford, H. Lee Woodcock, Yujin Wu, Charles L. Brooks, Bernard R. Brooks, Martin KarplusThe Journal of Physical Chemistry B 2024 · DOI

TL;DR

CHARMM, a cornerstone in computational biochemistry for nearly half a century, has undergone significant advancements in speed, accessibility, and functionality. This review details key developments since 2009, highlighting new GPU-accelerated engines, user-friendly interfaces, and a broader suite of simulation and analysis methods, ensuring its continued relevance for tackling complex biomolecular problems.

What Problem Does This Paper Address?

Computational biochemistry and biophysics rely heavily on molecular simulation software to understand complex biological systems. CHARMM, a leading program in this field, has been continuously evolving for decades. However, keeping pace with rapid advancements in computational hardware (especially GPUs) and the increasing complexity of biological questions requires constant updates. This paper addresses the need to document and summarize the major enhancements to CHARMM that have occurred since its last comprehensive review in 2009, ensuring researchers are aware of its expanded capabilities and can leverage them effectively.

Methods

This paper is a comprehensive review of the CHARMM software package, detailing its evolution and new features. Key advancements include:

  • GPU-Accelerated Engines: Integration of specialized GPU kernels and APIs (OpenMM, DOMDEC, BLADE, apoCHARMM) to significantly speed up molecular dynamics (MD) and related calculations.
  • Streamlined Workflows: Development of user-friendly interfaces and scripting tools (pyCHARMM, crimm, CHARMM-GUI) to simplify system preparation, simulation setup, and data analysis, making complex workflows more accessible.
  • Expanded Simulation and Analysis Methods: Introduction and refinement of a vast array of techniques, including:
  • Free Energy Methods: Enhanced alchemical path calculations, hybrid sampling, and improved convergence strategies.
  • Constant pH Methods: Advanced explicit and implicit solvent approaches, including proton hopping and replica exchange.
  • Enhanced Sampling and Transition Path Methods: String method, adaptively biased path optimization, and other techniques for exploring complex conformational landscapes.
  • QM/MM Methods: Integration of various quantum mechanical methods (SE, DFT, ab initio) with molecular mechanics for studying chemical reactions.
  • Advanced Energy Functions and Coarse Graining: Development of multipolar electrostatics, machine-learning-based energy functions, and transferable coarse-grained models.
  • Force Field Development: Continuous updates and expansions to CHARMM's force fields, including additive (CHARMM36) and polarizable (Drude) models for proteins, nucleic acids, lipids, and carbohydrates.

Datasets

Not explicitly discussed in the paper. The paper focuses on the software's capabilities and methods rather than specific benchmark datasets. However, it mentions that force field parameters are often developed and validated against experimental data and high-level quantum mechanical calculations.

Key Insights and Ideas

  • GPU Acceleration is Paramount: The integration of GPU-accelerated engines has been a major driver of CHARMM's recent progress, making computationally intensive simulations feasible within practical timeframes.
  • Accessibility is Key: The development of user-friendly interfaces like CHARMM-GUI and scripting capabilities through Python (pyCHARMM) is crucial for democratizing access to powerful simulation tools.
  • Modularity Enables Versatility: CHARMM's modular design allows for the seamless integration of diverse methods, from quantum mechanics to coarse-graining, and its extensibility through APIs.
  • Force Fields Continue to Evolve: The ongoing development of both additive and polarizable force fields is critical for accurately representing the complex interactions in biological systems.
  • QM/MM is Indispensable for Reactivity: For studying chemical reactions in biological contexts, QM/MM methods are essential, and CHARMM offers a comprehensive suite of tools for this purpose.

Key Results

  • Significant Speedups: GPU-accelerated engines provide performance comparable to or better than other leading simulation packages. Table 1 shows examples of ns/day performance on various systems and GPUs.
  • Broader Applicability: CHARMM now encompasses a vast array of methods, from atomistic to coarse-grained and QM/MM, enabling the study of a wider range of biomolecular phenomena.
  • Improved User Experience: New interfaces and scripting tools have made CHARMM more accessible and easier to use for complex workflows.
  • Enhanced Force Field Accuracy: Continuous refinement of force fields has led to better descriptions of molecular interactions and properties.

Limitations and Future Work

  • Drude Polarizable FF Access: Full access to the Drude polarizable force field through the CHARMM/OpenMM API is still under development.
  • DOMDEC Scalability: While DOMDEC offers parallel scalability, it is noted that apoCHARMM and other engines provide comparable performance on modern hardware.
  • Computational Cost of QM/MM: Despite advancements, QM/MM methods, especially those involving high-level quantum mechanics, remain computationally expensive, limiting routine use for extensive simulations.
  • Parameterization Challenges: Developing accurate force fields for complex molecules and systems (e.g., non-standard amino acids, modified nucleotides, ionic liquids) remains an ongoing effort.
  • Future Directions: The paper implicitly points to future work in areas such as further GPU acceleration, integration with machine learning, development of more robust QM/MM boundary treatments, and expansion of force field coverage. The continued availability of CHARMM for academic and nonprofit research is also highlighted as a key aspect for future development and adoption.

Summary updated

Abstract

Since its inception nearly a half century ago, CHARMM has been playing a central role in computational biochemistry and biophysics. Commensurate with the developments in experimental research and advances in computer hardware, the range of methods and applicability of CHARMM have also grown. This review summarizes major developments that occurred after 2009 when the last review of CHARMM was published. They include the following: new faster simulation engines, accessible user interfaces for convenient workflows, and a vast array of simulation and analysis methods that encompass quantum mechanical, atomistic, and coarse-grained levels, as well as extensive coverage of force fields. In addition to providing the current snapshot of the CHARMM development, this review may serve as a starting point for exploring relevant theories and computational methods for tackling contemporary and emerging problems in biomolecular systems. CHARMM is freely available for academic and nonprofit research at https://academiccharmm.org/program.
BibTeX
@article{hwang2024charmm,
  author = {Wonmuk Hwang and Steven L. Austin and Arnaud Blondel and Eric D. Boittier and Stefan Boresch and Matthias Buck and Joshua Buckner and Amedeo Caflisch and Hao-Ting Chang and Xi Cheng and Yeol Kyo Choi and Jhih‐Wei Chu and Michael F. Crowley and Qiang Cui and A. Damjanović and Yuqing Deng and M. Devereux and Xinqiang Ding and Michael Feig and Jiali Gao and David R. Glowacki and James E. Gonzales and Mehdi Bagheri Hamaneh and Edward Harder and Ryan L. Hayes and Jing Huang and Yandong Huang and Phillip S. Hudson and Wonpil Im and Shahidul M. Islam and Wei Jiang and Michael R. Jones and Silvan Käser and Fiona L. Kearns and Nathan R. Kern and Jeffery B. Klauda and Themis Lazaridis and Jinhyuk Lee and Justin A. Lemkul and Xiaorong Liu and Yun Luo and Alexander D. MacKerell and Dan Thomas Major and Markus Meuwly and Kwangho Nam and Lennart Nilsson and Victor Ovchinnikov and Emanuele Paci and Soohyung Park and Richard W. Pastor and Amanda R. Pittman and Carol Beth Post and Samarjeet Prasad and Jingzhi Pu and Yifei Qi and Thenmalarchelvi Rathinavelan and Daniel R. Roe and Benoı̂t Roux and Christopher N. Rowley and Jana Shen and Andrew C. Simmonett and Alexander J. Sodt and Kai Töpfer and Meenu Upadhyay and Arjan van der Vaart and Luis Itza Vazquez-Salazar and Richard M. Venable and Luke Warrensford and H. Lee Woodcock and Yujin Wu and Charles L. Brooks and Bernard R. Brooks and Martin Karplus},
  title = {{CHARMM at 45: Enhancements in Accessibility, Functionality, and Speed}},
  journal = {The Journal of Physical Chemistry B},
  year = {2024},
  doi = {10.1021/acs.jpcb.4c04100},
}
Figure