What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to accelerate the simulation process through enhanced sampling methods in molecular simulations, especially for complex biomolecular systems. Specifically, the paper explores how to use parametric time - lagged t - distributed stochastic neighbor embedding (ptltSNE) to design collective variables (CVs), thereby improving the efficiency of metadynamics simulations. ### Background and Problems 1. **High Computational Cost**: Molecular dynamics (MD) simulations are very computationally expensive because a large number of inter - atomic potentials need to be calculated at each step, which limits the time scale of the simulation, usually only reaching the nanosecond to microsecond level. 2. **Insufficient Sampling of Important States**: The nanosecond to microsecond time scale is often not sufficient to sample all important states of the system, so enhanced sampling methods need to be developed to accelerate the exploration of these states. 3. **Selecting Appropriate Collective Variables**: Enhanced sampling methods such as metadynamics rely on predefined collective variables (CVs), and the selection of these variables is crucial for the sampling efficiency, especially in complex biomolecular systems. ### Solutions 1. **Parametric Time - Lagged t - Distributed Stochastic Neighbor Embedding (ptltSNE)** - **Non - linear Dimensionality Reduction**: ptltSNE is a non - linear dimensionality reduction method that can better describe the non - linear motion of molecular systems. - **Time Lag**: By introducing a time lag, ptltSNE can emphasize slow - motion rather than fast but violent motion. - **Neural Network**: Use a neural network to calculate the low - dimensional embedding, making it possible to calculate the low - dimensional embedding of new sample structures and calculate the derivatives of collective variables with respect to atomic Cartesian coordinates. 2. **Application Examples** - **Trp - cage Folding**: In the paper, the folding and unfolding trajectories of Trp - cage (a small protein) are used as test data. Collective variables are designed through ptltSNE and metadynamics simulations are carried out. - **α - RMSD Collective Variable**: To accelerate the formation of α - helices, an α - RMSD collective variable is also added. ### Experimental Results 1. **Metadynamics Simulations** - Using two collective variables designed by ptltSNE for 1.5 microseconds of metadynamics simulations, no folding events were observed. - After adding the α - RMSD collective variable, one folding event was observed in 350 nanoseconds of metadynamics simulations. 2. **Parallel - Temperature Metadynamics Simulations** - Using collective variables designed by ptltSNE for 200 nanoseconds of parallel - temperature metadynamics simulations, 10 folding events were observed. - In contrast, the standard parallel - temperature simulation only observed two folding events within the same time. ### Conclusion By combining ptltSNE, metadynamics, and parallel - temperature metadynamics, the paper successfully accelerates the folding process of Trp - cage, demonstrating the potential of ptltSNE in designing collective variables, especially in the simulation of complex biomolecular systems.

Acceleration of Molecular Simulations by Parametric Time-Lagged tSNE Metadynamics

Metadynamics to Enhance Sampling in Biomolecular Simulations

Frequency Adaptive Metadynamics for the Calculation of Rare-Event Kinetics.

Enhanced Sampling in Molecular Dynamics Using Metadynamics, Replica-Exchange, and Temperature-Acceleration

Combining stochastic resetting with Metadynamics to speed-up molecular dynamics simulations

Transition rates, survival probabilities, and quality of bias from time-dependent biased simulations

A Review of Enhanced Sampling Approaches for Accelerated Molecular Dynamics

Timewarp: Transferable Acceleration of Molecular Dynamics by Learning Time-Coarsened Dynamics

Adaptable Parallel Acceleration Strategy For Dynamic Monte Carlo Simulations Of Polymerization With Microscopic Resolution

Enhanced sampling of chemical and biochemical reactions with metadynamics

Short-Time Infrequent Metadynamics for Improved Kinetics Inference

A variational conformational dynamics approach to the selection of collective variables in metadynamics

NNP/MM: Accelerating molecular dynamics simulations with machine learning potentials and molecular mechanic

Flexible Gaussian Accelerated Molecular Dynamics to Enhance Biological Sampling

NNP/MM: Accelerating Molecular Dynamics Simulations with Machine Learning Potentials and Molecular Mechanics

Metadynamics: a method to simulate rare events and reconstruct the free energy in biophysics, chemistry and material science

Resetting Metadynamics

Time-Lagged t-Distributed Stochastic Neighbor Embedding (t-SNE) of Molecular Simulation Trajectories

Kinetics-Optimized Enhanced Sampling Using Mean First Passage Times

Enhancing torsional sampling using fully adaptive simulated tempering