-
Notifications
You must be signed in to change notification settings - Fork 0
Installation
In order to use this difficulty prediction, you need RAxML-NG installed somewhere on your system. You can find the install instructions here.
The easiest way to install Pythia is to use the python package manager pip:
pip install git+https://github.com/tschuelia/PyPythia.git
Verify the correct installation by running pythia -h
.
You can again use pip for this and simply specify the tag you wish to install, e.g. for version 0.0.1
run:
pip install git+https://github.com/tschuelia/PyPythia.git@0.0.1
You can install Pythia from source if you want to explore the code or get the lastest development version. To do so run the following steps:
git clone https://github.com/tschuelia/PyPythia.git
cd PyPythia
pip install .
Verify the correct installation by running pythia -h
.
Most issues when installing Pythia seem to arise from broken or non-working LightGBM installations. If you encounter any such problem, and none of the following options help, please refer to the LightGBM installation instructions for your operating system and install LightGBM manually before repeating the Pythia installation as described above.
Installing on MacBooks with M1 chips caused some trouble for some users that seem to be caused by LightGBM's multiprocessing support. If you encounter any errors with the log pointing to LightGBM, the first thing you could try is to install LightGBM using homebrew:
brew install lightgbm
This might take a few minutes to finish. Once this ran successfully you can try to rerun the install instructions above.
If this does not solve your problem, you can try to install LightGBM manually using pip and disabling the multiprocessing:
pip install lightgbm --install-option=--nomp
and then rerun the installation of PyPythia.
If you are having trouble running pythia, you can also replace pythia
with python pypythia/prediction.py
. For this to work you need to install Pythia from source and you need to be in the PyPythia directory (which you should be after the installation).
If this does not work either, as of version 1.0.1, PyPythia includes a Python script that allows predictions from code without installing Pythia. See the Usage wiki page for more details.