Personal Precise Force Field for Intrinsically Disordered and Ordered Proteins Based on Deep Learning

Ji, Xiaofan; Líu, Hao; Zhang, Yangpeng; Chen, Hai-Feng

doi:10.1021/acs.jcim.2c01501

Cited by 12 publications

(16 citation statements)

References 74 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, to accurately characterize the structural properties of IDPs using MD simulations, it is essential to construct a well-defined physical model. To this end, our group has previously developed several force fields specifically designed for IDPs, including ff99IDPs, , ff14IDPs, ff14IDPSFF, CHARMM36IDPSFF, , OPLSIDPSFF, ff03CMAP, ESFF1, and PPFF1 . These force fields have demonstrated excellent accuracy in simulating the local conformational details of IDPs, such as J-coupling and chemical shifts .…”

Section: Introductionmentioning

confidence: 99%

Balanced Three-Point Water Model OPC3-B for Intrinsically Disordered and Ordered Proteins

Pan

Chen

2023

J. Chem. Theory Comput.

Self Cite

View full text Add to dashboard Cite

Intrinsically disordered proteins (IDPs) play a critical role in many biological processes. Due to the inherent structural flexibility of IDPs, experimental methods present significant challenges for sampling their conformational information at the atomic level. Therefore, molecular dynamics (MD) simulations have emerged as the primary tools for modeling IDPs whose accuracy depend on force field and water model. To enhance the accuracy of physical modeling of IDPs, several force fields have been developed. However, current water models lack precision and underestimate the interaction between water molecules and proteins. Here, we used Monte-Carlo re-weighting method to re-parameterize a three-point water model based on OPC3 for IDPs (named OPC3-B). We benchmarked the performance of OPC3-B compared with nine different water models for 10 IDPs and three ordered proteins. The results indicate that the performance of OPC3-B is better than other water models for both IDPs and ordered proteins. At the same time, OPC3-B possess the power of transferability with other force field to simulate IDPs. This newly developed water model can be used to insight into the research of sequence-disordered-function paradigm for IDPs.

show abstract

Section: Introductionmentioning

confidence: 99%

Balanced Three-Point Water Model OPC3-B for Intrinsically Disordered and Ordered Proteins

Pan

Chen

2023

J. Chem. Theory Comput.

Self Cite

View full text Add to dashboard Cite

show abstract

“…The activation function for full-connected layers was the rectified linear unit (ReLU). It was initially trained with the dataset containing 151,310 protein structures derived from the PDB database …”

Section: Methodsmentioning

confidence: 99%

“…The task for our neural network model is a hybrid task combing prediction for the region of dihedral including seven classes named αR, αL, α+, α’, proline (P), beta-sheet (B), and coil (C), respectively, same with the previous work, which divided into ( multi-classification problem ) and dihedral value ( regression problem ). To measure the performance of the neural network model, we used the accuracy for the multi-classification task and mean average error (MAE) for the regression task, which were calculated by eqs and : …”

Section: Methodsmentioning

confidence: 99%

“…It was initially trained with the dataset containing 151,310 protein structures derived from the PDB database. 51 Transfer Learning Model. Transfer learning is a commonly used method of further retraining a model pretrained on a larger dataset to adapt for another specific task with limited data.…”

Section: ■ Materials and Methodsmentioning

confidence: 99%

“…Transfer learning is a commonly used method of further retraining a model pretrained on a larger dataset to adapt for another specific task with limited data. 52 The pre-trained model obtained from Personal Precise Force Field (PPFF1) 51 was transferred to further train on the dataset obtained from the allosteric database (ASD). All of the network parameters including initial weight as well as the hyperparameters including number and depth of layers, learning rate, and dropout were kept the same with initial training.…”

Section: ■ Materials and Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Research and Evaluation of the Allosteric Protein-Specific Force Field Based on a Pre-Training Deep Learning Model

Cui

et al. 2023

J. Chem. Inf. Model.

Self Cite

View full text Add to dashboard Cite

Allosteric modulators are important regulation elements that bind the allosteric site beyond the active site, leading to the changes in dynamic and/or thermodynamic properties of the protein. Allosteric modulators have been a considerable interest as potential drugs with high selectivity and safety. However, current experimental methods have limitations to identify allosteric sites. Therefore, molecular dynamics simulation based on empirical force field becomes an important complement of experimental methods. Moreover, the precision and efficiency of current force fields need improvement. Deep learning and reweighting methods were used to train allosteric protein-specific precise force field (named APSF). Multiple allosteric proteins were used to evaluate the performance of APSF. The results indicate that APSF can capture different types of allosteric pockets and sample multiple energy-minimum reference conformations of allosteric proteins. At the same time, the efficiency of conformation sampling for APSF is higher than that for ff14SB. These findings confirm that the newly developed force field APSF can be effectively used to identify the allosteric pocket that can be further used to screen potential allosteric drugs based on these pockets.

show abstract

Application of computational approaches in biomembranes: From structure to function

Guo

Bao

et al. 2023

WIREs Comput Mol Sci

View full text Add to dashboard Cite

Biological membranes (biomembranes) are one of the most complicated structures that allow life to exist. Investigating their structure, dynamics, and function is crucial for advancing our knowledge of cellular mechanisms and developing novel therapeutic strategies. However, experimental investigation of many biomembrane phenomena is challenging due to their compositional and structural complexity, as well as the inherently multi‐scalar features. Computational approaches, particularly molecular dynamics (MD) simulations, have emerged as powerful tools for addressing the atomic details of biomembrane systems, driving breakthroughs in our understanding of biomembranes and their roles in cellular function. This review presents an overview of the latest advancements in related computational approaches, from force fields and model construction to MD simulations and trajectory analysis. We also discussed current hot research topics and challenges. Finally, we outline future directions, emphasizing the integration of force field development, enhanced sampling techniques, and data‐driven approaches to accelerate the growth of this field in the years to come. We aim to equip readers with an understanding of the promise and limitations of emerging computational technologies in biomembrane systems and offer valuable recommendations for future research endeavors.This article is categorized under: Structure and Mechanism > Computational Biochemistry and Biophysics Molecular and Statistical Mechanics > Molecular Dynamics and Monte‐Carlo Methods

show abstract

Personal Precise Force Field for Intrinsically Disordered and Ordered Proteins Based on Deep Learning

Cited by 12 publications

References 74 publications

Balanced Three-Point Water Model OPC3-B for Intrinsically Disordered and Ordered Proteins

Balanced Three-Point Water Model OPC3-B for Intrinsically Disordered and Ordered Proteins

Research and Evaluation of the Allosteric Protein-Specific Force Field Based on a Pre-Training Deep Learning Model

Application of computational approaches in biomembranes: From structure to function

Contact Info

Product

Resources

About