-
Notifications
You must be signed in to change notification settings - Fork 25
Python server for knime
The Python installation allows other software to execute Python code, for example to produce plots. In particular there is KNIME that makes extensive use of Python plots.
There are two options:
- install Python on local computer (local Python installation)
- install Python on another computer in the network (remote Python server)
- we highly recommend to use the Anaconda Python distribution
- download Anaconda and install it on your computer
- start the Anaconda Launcher and make sure you have a Python 2.x version installed
- check by running in the command line (Mac: Terminal; Windows: Command Prompt):
python --version
- set up the connection to KNIME by following instructions below.
- alternatively, install Python (2.x) and the packages listed below.
- install Anaconda (or Python 2.x) on a computer accessible in your network
- if necessary (Python 2.x) install packages listed below.
- download the latest stable mpicbg-python binaries from here and unpack the zip-file at the location of your choice.
- use any editor to change your chosen path to the pyserver folder in the pyserver.sh.
- use the command line to start the python server:
./pyserver.sh
The port number can be passed as an option to the script.
./pyserver.sh 1199
It is not supported for now to setup a Python-server under Windows. Hopefully this will come soon in the future…
Go to KNIME > Preferences > KNIME > Python Scripting, to tell KNIME who to talk to.
host: localhost (or whatever IP or DNS of computer running the remote Python-server) port: 1198
- to check the connection drag and drop a Python Snippet node and connect to some data
- just execute (no configuration needed) and if successful the Python-server connection is working
The Anaconda Python distribution comes already with a number of useful and necessary packages, like NumPy, SciPy, Matplotlib, … We recommend using the Anaconda distribution because you do not have to take care about installing necessary packages.
However one could also use any Python 2.x installation. To prepare your Python installation for all the tds-templates, you need to install the additional packages listed below.
These are the links to the templates and might need to be entered into the KNIME preferences:
Python-plot:
https://raw.githubusercontent.com/knime-mpicbg/scripting-templates/master/knime-scripting-templates/Python/figure-templates.txtPython-snippet:
https://raw.githubusercontent.com/knime-mpicbg/scripting-templates/master/knime-scripting-templates/Python/snippet-templates.txt
To use all templates install these packages in your Python installation.
- scipy and numpy
- matplotlib
- ordereddict (for some python versions)
The plot 'Venn Diagramm' depends on the external package 'matplotlib-venn'. These instructions should help you to install this package under Windows:
- run Anaconda Command Prompt under administrator rights (maybe it works also without admin rights!?)
- on command prompt change directory to anaconda program, eg: (may this is also not needed!?)
cd "C:\Program Files\Anaconda\"
- install package by running:
easy_install matplotlib_venn
- restart Anaconda or iPython and you should be able to import package using: import matplotlib_venn
- if that works that packages should be installed and the Venn plot-template should run as well