Allow gcaer to run GCAE from any location

[richelbilderbeek]

Depends on:

• [x] #9

Currently, we use only the models (i.e. neural network architecture, etc) supplied with GenoCAE, stored in the Singularity container.

However, there is need to create models on-the-fly.

The problem is that GCAE looks for the models relative to that script's location. To work on Bianca, however, I've put GCAE in a Singularity container, where that script is in /opt/gcae. This means I cannot add models on the fly, as the container will -rightfully!- prevent me from doing so.

After #9, run_gcae.py can be run from any location. However, it must be accompanied by the models needed. Allow gcaer to run GCAE from any location, by also installing/copying/creating those model files in the same relative paths.

[richelbilderbeek]

Hmm, maybe creating multiple models is easier (#15), putting this Issue in the freezer for now.

[richelbilderbeek]

Set to low priority, as there is no focus now on using on-the-fly models.