Introduction

Welcome to SSAGES, our new and shiny advanced sampling package.

Over the past several decades, molecular simulation has emerged as a powerful tool for investigating a wide range of physical phenomena. Molecular simulation is, in essence, a computational “microscope” whereby computers are used to “look at” the properties of a system that are difficult to observe or measure through traditional experimental setups. The comparison between simulations and the corresponding experimental systems can sometimes be challenging, usually due to factors such as the length and time scales explored. In simulation, a molecular model must have sufficient temporal and spatial accuracy to resolve the fastest time scales and shortest length scales within a system. Unfortunately, due to computational constraints, this detailed resolution has limited the length of time and number of particles that a model can simulate, typically simulating systems that are smaller than analogous experimental setups in laboratory environments for much shorter times than the duration of the experiments. However in recent years, advancements in computational processing power, including custom built computer architectures, have continued to increase the time and length scales accessible by molecular simulation, with current state-of-the-art simulations able to analyze systems for milliseconds (10-3s).

Another challenge arises from the difficulty in obtaining good statistics from molecular simulations. Thermal fluctuations dominate motion at the nano-scale and result in motion that appears random (i.e. Brownian), with no two molecular trajectories being identical. As a result, statistically meaningful averages are necessary in order to calculate thermodynamic and kinetic quantities of interest in these systems. An incredibly powerful thermodynamic quantity referred to as the relative free energy of a system can be calculated in this way. The relative free energy can characterize underlying system behavior in the presence of the thermal-induced random noise. Performing this necessary averaging within simulations is challenging. In essence, the requirement of averaging compounds the issue of time scales described previously; not only must long simulations be performed, but they must be performed a prohibitively large number of times in order to extract sufficient statistics. It is therefore necessary to develop efficient techniques to calculate meaningful averages from simulations.

Advanced sampling methods represent a class of simulation techniques that seek to improve this improper averaging and accelerate the extraction of useful properties (e.g. free energies, transition paths) from simulations. At the heart of all advanced sampling methods, is statistical mechanics, a field of physics that relates microscopic phenomena (i.e. the motion of particles) to macroscopic observables (e.g. temperature and pressure). By taking advantage of statistical mechanics, advanced sampling methods are used to apply a systematic bias to a simulation to speed convergence, and then mathematically remove this bias to extract the true underlying behavior. Throughout the past decade, advanced sampling methods have become wildly successful, and have now become an essential component in the toolbox of molecular simulation.

Despite the demonstrated utility of advanced sampling techinques, they have only been adopted by a fraction of the scientists working in the field. One explanation for this slow adoption is technical: advanced sampling methods are complicated, and not all research groups have the expertise required in order to implement these methods themselves. In the worst case, this leads to long stages of code development, possibly leading to unknown implementation errors or insufficient validation. Even in cases when advanced sampling methods are implemented, they are typically done so with a specific problem in mind and are custom-built for a certain model or application. This specificity necessitates modification of the custom-built advanced sampling code when studying new systems. This prevents the distribution of code between researches in the field. As a result, the same methods are implemented again and again by different members of the community. Sadly, in molecular simulation, it is quite common to “reinvent the wheel”.

SSAGES is an answer to this problem. SSAGES (Suite for Advanced Generalized Ensemble Simulations) is a free, open-source software package that allows users to easily apply advanced sampling techniques to any molecular system of interest. Simply put, SSAGES is a wrapper that converts a molecular simulation engine (e.g. LAMMPS, NAMD) into an advanced sampling engine. SSAGES contains a library of widely used enhanced sampling methods that can be used to calculate everything from free energies to transition pathways. Importantly, SSAGES works with many of the widely used simulation packages, and can simply be added on top of the simulations a researcher is already running. SSAGES is implemented in a highly modular way, and is easily extended to incorporate a new method or to modify an existing one and has been rigorously tested to ensure the accuracy of its calculations.

In short, SSAGES makes advanced sampling methods easy. We hope that it will do just that for your research.