Change Light Atom Selection Criteria From Name To Mass

[skfegan]

Summary

This pull request changes how we identify heavy/light atoms. We are now avoiding using the hydrogen atom name.

Changes

Change 1 Change atom selectors:

• For light atoms the selection string passed to MDAnalysis select_atoms is now "prop mass <= 1.1"
• For heavy atoms the selection string passed to MDAnalysis select_atoms is now "prop mass > 1.1"

Change 2 Change bead definitions

• When there is no heavy atom in a molecule, assign the whole molecule to one united atom bead

Impact

• This should now be able to handle dummy atoms (mass usually 0, but not named as hydrogen)
• It will work correctly as long as the masses are assigned correctly, regardless of how atom names or types are listed in the force field.
• Hydrogen molecules can be included.
• There is no risk of confusion between hydrogen and helium atom names.

[skfegan]

@jkalayan or @ioanaapapa Could you please check if this fixes errors caused by dummy atoms in the 4-point water model?

[harryswift01]

These all look like good changes. It looks as though the test coverage has dropped ever so slightly though, I think this is in relation to line 345 in the levels.py file where there is an append for all now. I think this would be the case of updating the test_get_beads_united_atom_level test and we would be back up to 100% but apart from that I would be happy to approve this.

[jkalayan]

Changes look great @skfegan, this works for a simulation containing 4-point waters and ensures dummy atoms of zero mass are excluded as united atoms.

[ioanaapapa]

Looks good to me! For a small organic molecule for 1 frame, compared to the result obtained with waterEntropy and the charge swapping python script, it seems to give a similar result.
SHF_CodeEntropy_light_atoms_1frame.txt
SHF_waterEntropy_swap.txt

[harryswift01]

These changes look great and it looks like we've now got a more robust way of ensuring we select the right atoms and tests have been added to ensure this works as expected, I'm very happy to approve this.