Problem with building from Dockerfile

[yann-zhong]

Hello,

I am currently trying to install TADbit by building from the Dockerfile within the containers folder.
To do so, I cloned the repo locally and created an empty folder, in which I ran the command:

docker build -f /mnt/d/Documents/Git/TADbit/containers/Dockerfile -t tadbit .

However, I run into an issue a few steps in, on the line:
RUN conda config --add channels salilab && conda config --add channels bioconda && \ conda install -y -q imp scipy matplotlib jupyter mcl samtools sra-tools pysam && \ conda clean -y --all && rm -rf /opt/conda/pkgs/*

The error, I believe, comes from the fact that the Dockerfile builds from the latest version of Python 2 (2.7), but that jupyter, scipy and matplotlib require Python3, as shown below:

I then tried to build from Python 3 (which I'm not even sure is a good idea) by changing the 10th line of the Dockerfile to get Miniconda3 instead of Miniconda2, but I'm faced with a similar error, this time with imp (which I understand is deprecated in favor of importlib) and scipy:

Would anyone have suggestions to help deal with these issues?
Thank you.

[david-castillo]

Hi,

Sorry for the delay.
We always have problems with the conda dependencies.
Can you try this line?

RUN conda config --add channels salilab && conda config --add channels bioconda && \ conda install -y -q python=3.7 imp scipy matplotlib jupyter mcl samtools sra-tools pysam -c conda-forge -c salilab -c bioconda && \ conda clean -y --all && rm -rf /opt/conda/pkgs/*

Regards

David