PRS directional response

[jamesmkrieger]

Here is a fix that I have been wanting for a long time. PRS now works to give directional responses using SignDy ModeEnsemble, giving n_atoms members with n_repeats modes each over 3 x n_atoms

[karolamik13]

Tests below:
In [1]: from prody import *
...:
...: atoms = parsePDB('1tqn', subset='calpha')
...:
...: anm = ANM('1tqn')
...: anm.buildHessian(atoms)
...: anm.calcModes(n_modes=20)
...:
...: prs = calcPerturbResponse(
...: anm,
...: atoms=atoms,
...: return_vectors=True,
...: repeats=5
...: )
...:
...: type(prs)
...: prs.numModeSets()
...: prs.numModes()
...: prs.numAtoms()
...:
@> PDB file is found in working directory (1tqn.pdb.gz).
@> 468 atoms and 1 coordinate set(s) were parsed in 0.01s.
@> Hessian was built in 0.21s.
@> 20 modes were calculated in 0.14s.
@> Calculating directional perturbation response [ 99%] 1sOut[1]: 468

In [2]: prs.numModeSets() == atoms.numAtoms()
...: prs.numModes() == 5
...: prs.numAtoms() == atoms.numAtoms()
...:
Out[2]: True

In [3]: prs_one = calcPerturbResponse(
...: anm,
...: atoms=atoms,
...: return_vectors=True,
...: perturb_node=10,
...: force=[1, 0, 0])
...:
...: prs_one.numModeSets()
...: prs_one.numModes()
...: prs_one.getLabels()
...:
@> Calculating directional perturbation response [ 0%]Out[3]: ['perturb_38']

In [4]: prs_one.numModeSets() == 1
...: prs_one.numModes() == 1
...:
Out[4]: True

In [5]: calcOverlap(prs_one.getModeSets()[0], anm[0])
Out[5]: array([0.69835304])

And test in ~/Prody:
$ python -m pytest -v $(git show --name-only --format="" 1278eda | grep 'test.*.py$')

========================================== test session starts ==========================================
platform linux -- Python 3.10.15, pytest-8.3.3, pluggy-1.5.0 -- /home/XXX/anaconda3/envs/prody_tests/bin/python
cachedir: .pytest_cache
rootdir: ~/ProDy
configfile: pyproject.toml
plugins: skip-slow-0.0.5, anyio-4.9.0
collected 7 items

prody/tests/dynamics/test_perturb.py::TestCalcPerturbResponse::testDirectionalPRSRecoversClosure PASSED [ 14%]
prody/tests/dynamics/test_perturb.py::TestCalcPerturbResponse::testEigvalsAreSquaredResponseMagnitude PASSED [ 28%]
prody/tests/dynamics/test_perturb.py::TestCalcPerturbResponse::testGNMRaisesForVectors PASSED [ 42%]
prody/tests/dynamics/test_perturb.py::TestCalcPerturbResponse::testLegacyReturnsTriple PASSED [ 57%]
prody/tests/dynamics/test_perturb.py::TestCalcPerturbResponse::testPerturbNodeList PASSED [ 71%]
prody/tests/dynamics/test_perturb.py::TestCalcPerturbResponse::testResponseVectorEqualsCovDotForce PASSED [ 85%]
prody/tests/dynamics/test_perturb.py::TestCalcPerturbResponse::testReturnVectorsEnsembleShape PASSED [100%]

=========================================== warnings summary ============================================
:241
:241: DeprecationWarning: builtin type SwigPyPacked has no module attribute

:241
:241: DeprecationWarning: builtin type SwigPyObject has no module attribute

:241
:241: DeprecationWarning: builtin type swigvarlink has no module attribute

prody/tests/dynamics/test_perturb.py::TestCalcPerturbResponse::testDirectionalPRSRecoversClosure
prody/tests/dynamics/test_perturb.py::TestCalcPerturbResponse::testDirectionalPRSRecoversClosure
~/ProDy/prody/utilities/eigtools.py:44: DeprecationWarning: Keyword argument 'turbo' is deprecated in favour of 'driver=gvd' keyword instead and will be removed in SciPy 1.12.0.
values, vectors = linalg.eigh(M, turbo=turbo, eigvals=eigvals)

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
===================================== 7 passed, 5 warnings in 2.01s =====================================

[jamesmkrieger]

Does it behave like you’d expect? It seems like pushing residues that are supposed to move is what has the biggest effect, which is against the idea of sensors and effectors. I think I want to try it out more and probably recode it again myself. The ordering in the file is messy anyway with it not reusing bits.

[jamesmkrieger]

It does behave as expected after trying more things but the code logic isn't great, so I am still making another version