specify dependecy versions

[chklovski]

specify dependecy versions

[nick-youngblut]

To deal with this issue for now, I just cleared out the dependencies set in setup.py. In other words: install_requires=(). This allowed me to install & run checkm2 successfully.

[chklovski]

Thanks for testing this. I've implemented your solution of removing requirements from the setup.py as this seems to resolve a lot of conflicts. Please re-open if issue surfaces again.

[nick-youngblut]

Which version of scipy was/is used for checkm2 development?

[nick-youngblut]

There are also potential conflicts between numpy and tensorflow, such as:

error: numpy 1.24.2 is installed but numpy~=1.19.2 is required by {'tensorflow'}

... when tensorflow is specified as tensorflow>=2.1.0,<2.6.0

I'm having a hard time finding a set of versions that satisfies python (3.9), numpy, grpcio, scipy, and tensorflow.

[nick-youngblut]

I tried the following env yaml:

channels:
  - conda-forge
  - bioconda
  - defaults
dependencies:
  - python=3.9
  - scikit-learn=0.23.2
  - h5py=2.10.0
  - numpy=1.19.2
  - diamond=2.0.4
  - tensorflow>=2.1.0,<2.6.0
  - lightgbm=3.2.1
  - pandas<=1.4.0
  - scipy=1.9
  - prodigal>=2.6.3
  - setuptools
  - requests
  - packaging
  - tqdm

...but this generates the error:

 grpcio 1.32.0 is installed but grpcio>=1.48.2 is required by {'tensorboard'}

...at least when using mambaorg/micromamba:1.1.0 as the base image in my Dockerfile.

[nick-youngblut]

If I try:

channels:
  - conda-forge
  - bioconda
  - defaults
dependencies:
  - python=3.9
  - grpcio
  - scikit-learn=0.23.2
  - h5py=2.10.0
  - numpy>=1.19.2
  - diamond=2.0.4
  - tensorflow>=2.1.0,<2.6.0
  - lightgbm=3.2.1
  - pandas<=1.4.0
  - scipy>=1.8
  - prodigal>=2.6.3
  - setuptools
  - requests
  - packaging
  - tqdm

...then I get:

error: numpy 1.24.2 is installed but numpy~=1.19.2 is required by {'tensorflow'}

[nick-youngblut]

I tried explicitly running the setup as in the README:

git clone --recursive https://github.com/chklovski/checkm2.git --branch 1.0.1
cd checkm2
mamba env create -f checkm2.yml 
python setup.py install

This results in the following error:

error: grpcio 1.32.0 is installed but grpcio>=1.48.2 is required by {'tensorboard'}

However, pip install "grpcio>=1.48.2" results in the following:

Requirement already satisfied: grpcio>=1.48.2 in /home/gitpod/.conda/envs/checkm2/lib/python3.8/site-packages (1.52.1)

...and python3.8/site-packages/ contains both grpcio-1.52.1.dist-info and grpcio-1.32.0-py3.8-linux-x86_64.egg.

mamba env create -f checkm2.yml is installing grpcio 1.52.1, so python setup.py install must be additionally installing grpcio-1.32.0.

I'm using Ubuntu 20.04.5 for this test.