Microbial Metabolic interactions
MMinte (pronounced /‘minti/) is a set of widgets that allows you to explore the pairwise interactions (positive or negative) that occur in a microbial community. From an association network and 16S rDNA sequence data, MMinte identifies corresponding genomes, reconstructs metabolic models, estimates growth under specific metabolic conditions, analyzes pairwise interactions, assigns interaction types to network links, and generates the corresponding network of interactions.
pip install mminte
Note that MMinte requires cobrapy 0.5.11. Python versions 2.7.11+, Python 3.5+, or Python 3.6+ are recommended.
Virtual environment installation
Here are step-by-step instructions for installing MMinte in a virtual environment.
Note on macOS El Capitan there is a known problem with the virtualenvwrapper dependency on the six package. For a work-around use this command:
$ pip install virtualenvwrapper –upgrade –ignore-installed six
Create a virtualenv for MMinte with these commands:
$ mkvirtualenv mminte
Use the --python option to select a specific version of Python for the virtualenv. For example, python=python3 to select the current python3 installed on the system.
Note on macOS, matplotlib requires Python be installed as a framework but virtualenv creates a non-framework build of Python. See the matplotlib FAQ for details on a workaround.
Upgrade pip and setuptools to the latest versions with this command:
(mminte)$ pip install --upgrade pip setuptools
Install the base MMinte package with this command:
(mminte)$ pip install mminte
If you want to verify the installation with the included tests, install the test dependencies and run the tests with these commands:
(mminte)$ pip install mminte[test] (mminte)$ pytest mminte -v
When you are done using MMinte, you can switch back to the system-installed version of Python with this command:
Start the MMinte virtualenv again with this command:
(mminte)$ workon mminte
How to run an analysis with MMinte
There are two ways to run an analysis with MMinte.
In a notebook
A tutorial of how to use MMinte is provided in a Jupyter notebook. Here’s how to start Jupyter and run the notebook from the virtual environment.
Install Jupyter with this command:
(mminte)$ pip install jupyter
Install a kernel that uses the virtualenv installation with this command:
(mminte)$ ipython kernel install --name "MMinte" --user
Start the Jupyter notebook server with this command:
(mminte)$ jupyter notebook
Jupyter opens a web page in your default browser with a file browser.
Navigate to the folder where MMinte is installed (use pip show mminte to find the folder) and in the “notebooks” folder click on the “tutorial.ipynb” notebook.
After the notebook opens, from the “Kernel” menu, select “Change kernel” and click on “MMinte”.
Now you can run the cells in the notebook.
In a web browser
MMinte includes a web site that allows you to run each widget from a web browser. The web site runs on your local system and requires that you use a folder on the local system for storing the results of the analysis. Here’s how to start the web site:
Install the additional packages needed for the web site interface with this command:
(mminte)$ pip install mminte[site]
Start the web server with this command:
(mminte)$ launchMMinte Logging to "/var/folders/pz/r04ddhtx6vgb48tg0dn5cys8vz00jn/T" folder [14/Apr/2017:14:25:04] ENGINE Listening for SIGHUP. [14/Apr/2017:14:25:04] ENGINE Listening for SIGTERM. [14/Apr/2017:14:25:04] ENGINE Listening for SIGUSR1. [14/Apr/2017:14:25:04] ENGINE Bus STARTING CherryPy Checker: The Application mounted at '' has an empty config. [14/Apr/2017:14:25:04] ENGINE Started monitor thread 'Autoreloader'. [14/Apr/2017:14:25:04] ENGINE Started monitor thread '_TimeoutMonitor'. [14/Apr/2017:14:25:04] ENGINE Serving on http://127.0.0.1:8080 [14/Apr/2017:14:25:04] ENGINE Bus STARTED
When you see the ENGINE Bus STARTED message, the web server is ready.
If another service is using port 8080, you can start the web server on a different port with this command:
(mminte)$ launchMMinte --port 8099
Open a web browser and go to http://localhost:8080 (change the port number if you started the web server on a different port) and follow the directions to run your analysis.
Version 1.0.3 (June 27, 2017)
- Added save_model_to_file() function
- Added apply_medium() function to default import
Version 1.0.2 (May 18, 2017)
- Updated handling of objectives for compatibility with cobra 0.6
- Fixed display of interaction network visualization when using different analysis folder in web site interface
Version 1.0.1 (May 5, 2017)
- Updated directions for virtual environment installation
- Updated version dependency for mackinac
- Fixed reading files with different newline characters
- Added notebooks to package distribution and updated directions
Version 1.0 (April 17, 2017)
Refactor of original MMinte with
- a simpler interface for functions,
- multiprocessing support for creating interaction models and calculating growth rates,
- updated web site that uses new version of DataSpyre,
- documentation in Jupyter notebooks,
- a test suite,
- reorganized repository to enable installation with pip.
How to cite MMinte
If you use MMinte for an analysis, please cite this paper: MMinte: an application for predicting metabolic interactions among the microbial species in a community
- The models provided in the mminte/test/data folder are from Anoxic Conditions Promote Species-Specific Mutualism between Gut Microbes In Silico.
- The 16S sequences included in the database were provided by Maulik Shukla on the 3rd of November of 2015.