Skip to content
Switch branches/tags

Latest commit

* Add method to flatten an mbuild compound

* add some unit tests

* fix typos

* add inplace option for flatten method

* fix typo in test

* fix typo

* properly add ports in Compound.flatten and add tests

* add test for inplace=False

* minor change to is_independent to improve performance

Git stats


Failed to load latest commit information.
Latest commit message
Commit time

mBuild: a hierarchical, component based molecule builder

Gitter chat CI Anaconda Badge codecov

With just a few lines of mBuild code, you can assemble reusable components into complex molecular systems for molecular dynamics simulations.

  • mBuild is designed to minimize or even eliminate the need to explicitly translate and orient components when building systems: you simply tell it to connect two pieces!
  • mBuild keeps track of the system's topology so you don't have to worry about manually defining bonds when constructing chemically bonded structures from smaller components.

To learn more, get started or contribute, check out our website.

mBuild within the MoSDeF Ecosystem

mBuild within the MoSDeF Ecosystem

The mBuild package is part of the Molecular Simulation Design Framework (MoSDeF) project. Libraries in the MoSDeF ecosystem are designed to provide utilities neccessary to streamline a researcher's simulation workflow. When setting up simulation studies, we also recommend users to follow the TRUE (Transparent, Reproducible, Usable-by-others, and Extensible) standard, which is a set of common practices meant to improve the reproducibility of computational simulation research.

Quick Start with Docker

To use mbuild in a jupyter-notebook that runs from a docker container with all the dependencies installed use the following command:

$ docker pull mosdef/mbuild:latest
$ docker run -it --name mbuild -p 8888:8888 mosdef/mbuild:latest su anaconda -s\
      /bin/sh -l -c "jupyter-notebook --no-browser --ip="" --notebook-dir\

Alternatively, you can also start a Bourne shell directly:

$ docker run -it --name mbuild mosdef/mbuild:latest

To learn more about using mBuild with docker, please refer to the documentation here.


Interactive tutorials can be found here:


Components in dashed boxes are drawn by hand using, e.g., Avogadro or generated elsewhere. Each component is wrapped as a simple python class with user defined attachment sites, or ports. That's the "hard" part! Now mBuild can do the rest. Each component further down the hierarchy is, again, a simple python class that describes which piece should connect to which piece.

Ultimately, complex structures can be created with just a line or two of code. Additionally, this approach seamlessly exposes tunable parameters within the hierarchy so you can actually create whole families of structures simply by adjusting a variable:

import mbuild as mb
from mbuild.examples import PMPCLayer

pattern = mb.Random2DPattern(20)  # A random arrangement of 20 pieces on a 2D surface.
pmpc_layer = PMPCLayer(chain_length=20, pattern=pattern, tile_x=3, tile_y=2)

Zwitterionic brushes on beta-cristobalite substrate

Community Recipes

Use case-specific systems can be generated via mBuild recipes. Some users have graciously contributed recipes for particular systems, including:

Citing mBuild

If you use this package, please cite our paper. The BibTeX reference is

      author = "Klein, Christoph and Sallai, János and Jones, Trevor J. and Iacovella, Christopher R. and McCabe, Clare and Cummings, Peter T.",
      title = "A Hierarchical, Component Based Approach to Screening Properties of Soft Matter",
      booktitle = "Foundations of Molecular Modeling and Simulation",
      series = "Molecular Modeling and Simulation: Applications and Perspectives",
      year = "2016",
      doi = ""


Various sub-portions of this library may be independently distributed under different licenses. See those files for their specific terms.

This material is based upon work supported by the National Science Foundation under grants NSF CBET-1028374 and NSF ACI-1047828. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.