Have CLI place generated YAML file into the perses calculation directory

[jchodera]

Currently, the CLI generates a YAML file with the updated overridden fields, but it places it into a file like

parsed-2022-04-18-20:51:55-my.yaml

rather than actually putting it in the perses calculation output directory where it would be useful.

We should fix this so that we can pull useful information from it, like what calculation was actually performed to generate all the files in this directory.

[jchodera]

To be completely clear, I'm recommending that instead of this:

├── parsed-2022-04-18-20:51:55-my.yaml <--- it's currently here
├── long_run_lig0to8
│   ├── atom_mapping.png
│   ├── out-complex_checkpoint.nc
│   ├── out-complex.nc
│   ├── out-complex.pdb
│   ├── out-complex_real_time_analysis.yaml
│   ├── out-hybrid_factory.npy.npz
│   ├── out-solvent_checkpoint.nc
│   ├── out-solvent.nc
│   ├── out-solvent.pdb
│   ├── out-solvent_real_time_analysis.yaml
│   ├── out-topology_proposals.pkl
│   └── xml
│       ├── complex-hybrid-system.gz
│       ├── complex-new-system.gz
│       ├── complex-old-system.gz
│       ├── solvent-hybrid-system.gz
│       ├── solvent-new-system.gz
│       └── solvent-old-system.gz

we want something like this:

├── long_run_lig0to8
│   ├── perses-cli.yaml <----- we want it here
│   ├── atom_mapping.png
│   ├── out-complex_checkpoint.nc
│   ├── out-complex.nc
│   ├── out-complex.pdb
│   ├── out-complex_real_time_analysis.yaml
│   ├── out-hybrid_factory.npy.npz
│   ├── out-solvent_checkpoint.nc
│   ├── out-solvent.nc
│   ├── out-solvent.pdb
│   ├── out-solvent_real_time_analysis.yaml
│   ├── out-topology_proposals.pkl
│   └── xml
│       ├── complex-hybrid-system.gz
│       ├── complex-new-system.gz
│       ├── complex-old-system.gz
│       ├── solvent-hybrid-system.gz
│       ├── solvent-new-system.gz
│       └── solvent-old-system.gz

[jchodera]

We could store the metadata currently stored in the filename (such as timestamp the file was generated) inside the YAML file it generates.

[jchodera]

It would also be useful to store the molecule titles (molecule names) from the SDF file corresponding to the initial and final ligands, since we need to use these in constructing the overall transformation network for analysis and plotting.

[mikemhenry]

We have moved the parsed yaml into the calculation directory but I'm going to keep this issue to track:

• store the metadata currently stored in the filename (such as timestamp the file was generated) inside the YAML file
• store the molecule titles (molecule names) from the SDF file corresponding to the initial and final ligands

I think that it makes to add basically two sections to the yaml, one section that has all the parsed options/defaults, and another section that has metadata, like the GPU used, the time created, hostname the system was ran on, ligand name (not just index) stuff like that.

[ijpulidos]

The first part is solved via #972. I'm creating a new issue with the second part for the molecules titles.

[ijpulidos]

Oh my bad, the tab didn't update itself and I missed @mikemhenry comment. Sorry! Will reopen.

[mikemhenry]

No worries, when I saw it was closed I thought "wow I can't believe Iván got those two things done already"