Adding LambdaSettings object for AFE

[hannahbaumann]

Developers certificate of origin

• I certify that this contribution is covered by the MIT License here and the Developer Certificate of Origin at https://developercertificate.org/.

[pep8speaks]

Hello @hannahbaumann! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file openfe/protocols/openmm_afe/equil_afe_settings.py:

Line 67:60: W291 trailing whitespace
Line 69:80: E501 line too long (82 > 79 characters)
Line 78:80: E501 line too long (83 > 79 characters)
Line 148:78: W291 trailing whitespace

• In the file openfe/protocols/openmm_afe/equil_solvation_afe_method.py:

Line 546:80: E501 line too long (81 > 79 characters)
Line 547:80: E501 line too long (80 > 79 characters)
Line 562:42: E741 ambiguous variable name 'l'

• In the file openfe/protocols/openmm_rfe/equil_rfe_settings.py:

Line 39:1: W293 blank line contains whitespace
Line 40:74: W291 trailing whitespace

• In the file openfe/tests/protocols/test_openmm_afe_slow.py:

Line 62:9: E128 continuation line under-indented for visual indent
Line 65:9: E128 continuation line under-indented for visual indent

• In the file openfe/tests/protocols/test_openmm_afe_solvation_protocol.py:

Line 50:80: E501 line too long (85 > 79 characters)

Comment last updated at 2024-02-01 17:40:36 UTC

[IAlibay]

@dwhswenson we're not 100% sure how the settings docs is meant to work - do you have a better idea here? We can add the relevant autopydantic entry under docs, but it looks like AlchemicalSettings isn't behaving as expected even when there is such an entry.

[dwhswenson]

@IAlibay : could you clarify on the settings docs issue? I don't see anything wrong.

This PR's docs build for AlchemicalSettings has no field for lambda_functions/lambda_windows, whereas they do exist in current stable docs. Isn't that the change you would expect?

You'll see that this PR's build doesn't have a link for LambdaSettings, but that should just require adding the .. autopydantic_model:: at docs/reference/api/openmm_rfe.rst, etc.

[IAlibay]

@IAlibay : could you clarify on the settings docs issue? I don't see anything wrong.

This PR's docs build for AlchemicalSettings has no field for lambda_functions/lambda_windows, whereas they do exist in current stable docs. Isn't that the change you would expect?

You'll see that this PR's build doesn't have a link for LambdaSettings, but that should just require adding the .. autopydantic_model:: at docs/reference/api/openmm_rfe.rst, etc.

Let's add it to our topics of discussion later today - there's a couple of things I'm not 100% sure but it'll be easier to go through in person.

[IAlibay]

Validators discussed at today's call:

• naked charges
• monotonically increasing values

(for AFE)

[hannahbaumann]

There may be a better way to compare different list lengths here?

[hannahbaumann]

This validator currently would lead to errors when changing the settings from the default settings (since when changing lambda_elec the list of lambda_vdw has still the old length and therefore would raise an error. One could move the validator into the protocol instead, or do you have other suggestions @richardjgowers ?

[richardjgowers]

Yeah there might be a limit to what's possible in the @validator and instead you want a protocol function. The nice thing about these (when possible) is it fails fast & early.

[hannahbaumann]

I moved the check to the _get_lambda_schedule function, or would it be better to have a separate _validator function there?

[IAlibay]

Regarding our discussion on documentation yesterday, could you add autopydantic entries for LambdaSettings here https://github.com/OpenFreeEnergy/openfe/blob/main/docs/reference/api/openmm_rfe.rst and https://github.com/OpenFreeEnergy/openfe/blob/main/docs/reference/api/openmm_solvation_afe.rst please?

[IAlibay]

Couple of things here; some docs and a couple of validation things. Otherwise this is looking good!

[IAlibay]

Would it be reasonable here to change the default to match what we usually run as a default when doing AHFEs, i.e. 14 windows?

[hannahbaumann]

Ah yes, had changed that in the other PR, but not here, added this now!

[IAlibay]

To improve method reproducibility, I would switch this to passing self.settings.lambda_settings instead. That way when we make derived classes we can import the staticmethod across to different places and not face typing errors because we're using different protocol settings.

[IAlibay]

We should also check that there are any non-zero restraints and raise a logger warning if there are (i.e. this protocol doesn't apply restraints, restraints won't be applied).

[IAlibay]

Maybe add more details about what the settings class does?

Defines lists of floats to control various aspects of the alchemical transformation.

Notes
--------
* In all cases a lambda value of 0 defines a fully interacting state A and a non-interacting state B, whilst a value of 1 defines a fully interacting state B and a non-interacting state A.
* ``lambda_elec``, `lambda_vdw``, and ``lambda_restraints`` must all be of the same length, defining all the windows of the transformation.

[IAlibay]

Suggested change

	This describes the lambda schedule.
	Lambda schedule settings.
	Settings controlling the lambda schedule, these include the switching function type, and the number of windows.

Or something like that.

[IAlibay]

This is amazing, thanks!