(657b) Mosdef Cassandra: A Python Interface Providing Complete Integration between the Cassandra Monte Carlo Software and Mosdef Tools | AIChE

(657b) Mosdef Cassandra: A Python Interface Providing Complete Integration between the Cassandra Monte Carlo Software and Mosdef Tools


DeFever, R. S. - Presenter, Clemson University
Maginn, E. - Presenter, University of Notre Dame
Over the previous half-century, molecular simulations have proven an invaluable technique for investigating and predicting the behavior of matter. The Monte Carlo (MC) methods are particularly valuable for their ability to easily simulate open ensembles, i.e., including molecule insertions and deletions. This capability enables the simulation of phase equilibria, such as vapor-liquid equilibrium and adsorption in porous materials. Rapidly increasing computing power makes MC simulation, in principle, more powerful. However, generating input files and managing the overall simulation workflow still requires human intervention and suffers from human error. The next generation of simulation studies will likely include more high-throughput screening and perhaps even iterative approaches that combine statistical methods (e.g., machine learning) and molecular simulations to probe the behavior of chemical systems across large parameter spaces. These types of studies will require two key components from the next generation of MC simulation software: enhanced simulation reproducibility and improved integration with other community-based software tools.

In this work, we demonstrate our recent integration of the Molecular Simulation Design Framework (MoSDeF) [1] with Cassandra [2] via a new Python interface for Cassandra. MoSDeF is an effort to promote TRUE simulation workflows (Transparent, Reproducible, Usable by others and Extensible) by providing modular Python-based tools that can be used during the system creation and force field application steps of a simulation workflow. The goal of these tools is to remove the use of graphical user interfaces, in-house interconversion scripts, and other ad-hoc steps used by many researchers to create and modify input files for molecular simulations. The Python-based Cassandra interface offers three main advantages: improved reproducibility by containing the entire simulation workflow within a single script, a tremendously simplified process of performing an MC simulation, and easier integration with other software tools.

We describe the philosophy and design of the MoSDeF Cassandra interface and demonstrate how the software simplifies the act of performing an MC simulation. Starting from the “traditional” Cassandra simulation workflow, we show how MoSDeF Cassandra removes ad-hoc modifications and the need for graphical user interfaces. Instead, the entire simulation workflow (system creation, force field assignment, simulation, and analysis) is contained in a single Python script. This script is easy to version-control and share. Reproducing a simulation is as trivial as downloading the script and installing the appropriate software. Modifying the simulation (e.g., different conditions, molecules, etc.) is often nearly as trivial. We then demonstrate extensions to more complex workflows, such as calculating adsorption isotherms in porous media. Finally, we show how MoSDeF Cassandra integration with Signac [3] enables simplified job management and submission for thousands or more simulations. With this power, we believe MoSDeF Cassandra is well-positioned to enable the next generation of Monte Carlo simulations.

[1] M.W. Thompson, J.B. Gilmer, R.A. Matsumoto, C.D. Quach, P. Shamaprasad, A.H. Yang, C.R. Iacovella, C. McCabe, P.T. Cummings. Towards molecular simulations that are transparent, reproducible, usable by others, and extensible. Mol. Sim. In press, (2020).

[2] J.K. Shah, E. Marin-Rimoldi, R.G. Mullen, B.P. Keene, S. Khan, A.S. Paluch, N. Rai, L.L. Romanelo, T.W. Rosch, B. Yoo, E.J. Maginn, Cassandra: An open source Monte Carlo package for molecular simulation. J. Comput. Chem. 38 (2017) 1727-1739.

[3] V. Ramasubramani, C.S. Ardorf, P.M. Dodd, B.D. Dice, S.C. Glotzer, Simple data and workflow management with the signac framework. Comput. Mater. Sci. 146 (2018) 220-229.