Installation¶
Using environment.yml file¶
Currently, the fastest and easiest way to install dogsled is via Conda using the YML file located in the dogsled GitHub repository. For Linux and macOS:
user@arch:~$ git clone https://github.com/RhDm/dogsled.git
user@arch:~$ cd dogsled
user@arch:~$ conda env create -f environment.yml
user@arch:~$ conda activate dogsled
for Windows in the same way but with a different YML file:
user@arch:~$ conda env create -f win_environment.yml
This method installs autocamically all dependencies including libvips.
Via pip¶
dogsled can be installed also via pip (be sure to have Python 3.9 on your system and create a separate environment using, for example, virtuelenv):
user@arch:~$ pip install virtualenv
user@arch:~$ virtualenv /PATH/WHERE/PYTHON/ENV/WILL/LIVE --python=python3.9
user@arch:~$ source /PATH/WHERE/PYTHON/ENV/WILL/LIVE/bin/activate
user@arch:~$ pip install dogsled
This, however, will not install libvips on Linux and macOS, please follow the installation guides.
As dogsled uses Bayer’s paquo library for interactions with QuPath files, QuPath installation is also required if you install dogsled via pip (QuPath is automatically installed if Conda is used)
In case you plan to work with large slides (over 100,000pixels per side), or if your system can not handle
smaller slide sizes causing a crash, you might want to set vips_stitcher key value of the DEFAULTS
dictionary to True, see API for further details.