Feature/fasthydromap command

[samlobe]

Summary

This PR adds a new ChimeraX bundle for FastHydroMap (a new hydrophobicity software; see preprint) with two commands:

• fasthydromap install
• fasthydromap

The bundle installs FastHydroMap into a ChimeraX-managed virtual environment from PyPI and runs predictions on atomic structures from within ChimeraX.

Features

• managed install via fasthydromap install
• default CPU Torch install
• prediction quantities:
  • fdewet
  • pc1
  • pc2
  • pc3
• optional palette and range overrides for coloring
• no changes to the existing Hydrophobicity button or shortcuts

Example commands

• fasthydromap install
• fasthydromap #1
• fasthydromap #1 quantity pc1
• fasthydromap #1 palette ^lipophilicity range 3,7

Notes

This PR is intentionally command-only to keep the integration minimal and self-contained. If this FastHydroMap model becomes widely adopted, perhaps later the Hydrophobicity GUI button can be replaced with a call to this command, since the current MLP Hydrophobicity default is quite old (from 1998) and the hydrophobicity literature has advanced substantially since then. FastHydroMap is open-sourced and pip-installable.

[tomgoddard]

Thanks for making the fasthydromap ChimeraX command. Have you considered submitting this to the ChimeraX Toolshed (https://cxtoolshed.rbvi.ucsf.edu/) so that users can install it from within ChimeraX using menu Tools / More Tools? I am on vacation today so I haven't carefully studied how you install Torch and other dependencies but it looks similar to how ChimeraX installs Boltz in its own virtual environment. This can be slow to install since Torch (even CPU version) and other dependencies (e.g. Pandas) are large, so I'm not sure whether that should replace the ChimeraX MLP which is built-in to ChimeraX and is very light weight. But for users who want potentially more accurate results it is of course nice to have. We will take a look at your preprint to figure out if we want to put fasthydromap (and its install) directly into ChimeraX distributions instead of Toolshed. That is more convenient but usually that puts the UCSF team in the position of maintaining it, so we generally prefer that contributions go into the Toolshed so the contributors continue to be the maintainers of their code. Could you tell us whether you had in mind maintaining the ChimeraX fasthydromap and whether you think the ChimeraX Toolshed bundle submission works for you?