Interesting issue when running in conda environment

[tijeco]

As mentione in #4, I've installed mxnorm on a fresh computer in a conda environment.

I was going through the examples in the README and ran into this when I executed the follwowing:

mx_otsu = run_otsu_discordance(mx_norm,
                        table="both",
                        threshold_override = NULL,
                        plot_out = FALSE)

The output was:

No non-system installation of Python could be found.
Would you like to download and install Miniconda?
Miniconda is an open source environment management system for Python.
See https://docs.conda.io/en/latest/miniconda.html for more details.

Would you like to install Miniconda? [Y/n]

For context, outside of conda, I don't have python installed (there is no /usr/bin/python).

In the conda environment that I am using and have mxnorm installed in, python is installed, and should be accessible through the $PATH.

Also, I already have miniconda installed, so I certainly wouldn't want to proceed with installing it again.

I think some of this should be mentioned in the documentation, as far as dependencies go.

Also, is there a way to determine that miniconda is installed, or that mxnorm is already installed in a conda environment?

[ColemanRHarris]

As discussed in #4, I was able to setup a similar Docker container and re-create this issue. Based on some digging in the conda repository, this was a known issue (rstudio/reticulate#607) and was seemingly addressed. The steps I took to work around it were:

• Did not install Miniconda via the R console.
• Used reticulate::py_config() to setup my desired Python environment.
• Ensured scikit-image was installed in the same location as the Python environment.

These steps allowed me to run the command you listed above without error, and have been recreated in the Dependencies section of the README.