Dependency issue upon installation

[snixon2]

There are many dependency issue when running the following command on release 2.0.

source macos_installation.sh

Particularly, I suspect that openmmforcefields=0.11.2 and beliveau-lab::nupack=4 on e2edna-env.yml are problematic. When running the script (or simply the commands within it), it does not even make the e2edna conda environment.

I have tried both the working 2.0 version and this newest version. The old version (with the long .yml file) will have the version issues that can somewhat be fixed by pip installing a handful of lines. Still, it will result in MANY specifications which are incompatible with my system.

[taoliu032]

We noticed that "beliveau-lab::nupack=4" is only an available conda package, unsure why it happened. The way to work around is to download nupack installer wheel from its official website (free for academics), then install it directly in your created conda environment. We are unware of the openmmforcefields problems. Using a new version of the package may help.

[snixon2]

Thank you for your quick response. I am actively testing the newer .yml file and several packages are taking a very long time (>40 mins each) to install individually. I am trying again by modifying the script to do the following:

conda create -n e2edna

conda activate e2edna -y

conda env update -f e2edna-env.yml --prune.

In the e2edna-env.yml file, I am commenting out a few lines. I will update you if I get the installation working.

[taoliu032]

My experience is that it's possible that solving the package can take a long time, but the installation of package should be relatively faster. But sure, let us know :)

[snixon2]

When I did the process I mentioned, everything seems to have gone smoothly (including all the dependency issues). I guess manually making, activating, and updating a conda environment works instead of doing all at once, even though they should theoretically do the same thing.

[InfluenceFunctional]

In our experience trying it all at once often leads indeed to extremely long solve times. Quite annoying.