Unfortunately, we didn't have time for a full reproduction & review, but I wanted to drop a quick note here & share some appreciation for all the work you've put in the package!
One thing that got us stuck early on was getting things installed. The binder approach isn't working for us anymore, likely related to the Python version:
The suspicious part to us was this log message:
"Python version unspecified, using current default Python version 3.10. This will change in the future.Building conda environment for python=3.10"
We then tried to install it via the environment.yml file you provide, but that also made issues (hard to set up Python 3.7 envs on Apple Silicon macs & getting all the old fixed package versions.
In the end, we managed to install it just via pip install pykneer and could run the preprocessing notebooks. And I set up a conda env with relaxed dependencies (in Python 3.10) and we reproduced the plots of figure 4 with your notebook* :) 🎉
*) mostly reproduced. The figure in the notebook (even hosted on Github) is not a perfect 1-to-1 match, e.g. minor rounding differences of the values. But it looks very close!
Hey @sbonaretti
Super cool package & nice paper! We were looking at it today during the a repo-hack: https://www.reprohack.org/event/28/
Unfortunately, we didn't have time for a full reproduction & review, but I wanted to drop a quick note here & share some appreciation for all the work you've put in the package!
One thing that got us stuck early on was getting things installed. The binder approach isn't working for us anymore, likely related to the Python version:
The suspicious part to us was this log message: "Python version unspecified, using current default Python version 3.10. This will change in the future.Building conda environment for python=3.10"
We then tried to install it via the environment.yml file you provide, but that also made issues (hard to set up Python 3.7 envs on Apple Silicon macs & getting all the old fixed package versions.
In the end, we managed to install it just via
pip install pykneerand could run the preprocessing notebooks. And I set up a conda env with relaxed dependencies (in Python 3.10) and we reproduced the plots of figure 4 with your notebook* :) 🎉*) mostly reproduced. The figure in the notebook (even hosted on Github) is not a perfect 1-to-1 match, e.g. minor rounding differences of the values. But it looks very close!