Skip to content

atm #

Automated BFE calculations using the alchemical transfer method

Classes:

  • ATMSoftCore –

    Configuration for the ATM soft-core potential.

  • ATMAlignmentRestraint –

    Configuration for an ATM alignment restraint.

  • ATMRestraints –

    Configure the restraints that will be applied during the ATM calculations.

  • ATMReferenceSelection –

    Configure how receptor binding sites and ligand alignment reference atoms are

  • ATMSetupStage –

    Configure how the complex will be solvated and restrained prior to

  • ATMStates –

    Configure the lambda schedules.

  • ATMEquilibrateStage –

    Configure how the system will be equilibrated prior to replica exchange.

  • ATMSamplingStage –

    Configure how the system will be sampled using Hamiltonian replica exchange.

  • ATMConfig –

    Configuration the stages of the ATM calculation.

  • ATMEdge –

    Defines an ATM specific edge in a free energy network.

  • ATMNetwork –

    Defines an ATM specific free energy network.

Functions:

  • load_config –

    Load a configuration from a YAML file.

  • create_state_dicts –

    Map the lambda states specified in the configuration to a list of dictionaries.

  • compute_ddg –

    Computes the total binding free energy.

  • equilibrate_states –

    Equilibrate the system at each lambda window.

  • run_hremd –

    Perform replica exchange sampling for a system prepared for ATM calculations.

  • setup_system –

    Prepares a system ready for running the ATM method.

  • select_displacement –

    Attempts to automatically select a displacement vector for the ligands.

  • run_workflow –

    Run the setup, equilibration, and sampling phases.

  • submit_network –

    Submits a set of ATM calculations to the SLURM queueing manager.

Attributes:

DEFAULT_LAMBDA_1 module-attribute #

DEFAULT_LAMBDA_1 = (
    0.0,
    0.0,
    0.0,
    0.0,
    0.0,
    0.0,
    0.1,
    0.2,
    0.3,
    0.4,
    0.5,
    0.5,
    0.4,
    0.3,
    0.2,
    0.1,
    0.0,
    0.0,
    0.0,
    0.0,
    0.0,
    0.0,
)

The default lambda 1 schedule.

DEFAULT_LAMBDA_2 module-attribute #

DEFAULT_LAMBDA_2 = (
    0.0,
    0.1,
    0.2,
    0.3,
    0.4,
    0.5,
    0.5,
    0.5,
    0.5,
    0.5,
    0.5,
    0.5,
    0.5,
    0.5,
    0.5,
    0.5,
    0.5,
    0.4,
    0.3,
    0.2,
    0.1,
    0.0,
)

The default lambda 2 schedule.

DEFAULT_DIRECTION module-attribute #

DEFAULT_DIRECTION = (
    1,
    1,
    1,
    1,
    1,
    1,
    1,
    1,
    1,
    1,
    1,
    -1,
    -1,
    -1,
    -1,
    -1,
    -1,
    -1,
    -1,
    -1,
    -1,
    -1,
)

The default direction schedule.

DEFAULT_ALPHA module-attribute #

DEFAULT_ALPHA = (
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
    0.1,
)

The default alpha schedule.

DEFAULT_U0 module-attribute #

DEFAULT_U0 = (
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
    110.0,
)

The default u0 schedule.

DEFAULT_W0 module-attribute #

DEFAULT_W0 = (
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
)

The default w0 schedule.

DEFAULT_MAX_REST_TEMPERATURE module-attribute #

DEFAULT_MAX_REST_TEMPERATURE = 900.0

The default maximum temperature to use during the REST2 calculations.

DEFAULT_BM_B0 module-attribute #

DEFAULT_BM_B0 = tuple(tolist())

The default beta scaling factors to use if running with REST2.

DEFAULT_EQUILIBRATE_INTEGRATOR module-attribute #

DEFAULT_EQUILIBRATE_INTEGRATOR = LangevinIntegrator(
    timestep=2.0 * femtosecond, friction=1.0 / picosecond
)

The default integrator to use during equilibration.

DEFAULT_EQUILIBRATE_RESTRAINTS module-attribute #

DEFAULT_EQUILIBRATE_RESTRAINTS = {
    DEFAULT_RESTRAINT_MASK: FlatBottomRestraint(
        k=25.0 * _KCAL_PER_ANG_SQR, radius=1.5 * _ANGSTROM
    )
}

The default position restraints to apply during equilibration.

ATMSoftCore pydantic-model #

Bases: BaseModel

Configuration for the ATM soft-core potential.

ATMAlignmentRestraint pydantic-model #

Bases: BaseModel

Configuration for an ATM alignment restraint.

ATMRestraints pydantic-model #

Bases: BaseModel

Configure the restraints that will be applied during the ATM calculations.

ATMReferenceSelection pydantic-model #

Bases: BaseModel

Configure how receptor binding sites and ligand alignment reference atoms are selected if the user does not explicitly provide them.

ATMSetupStage pydantic-model #

Bases: BaseModel

Configure how the complex will be solvated and restrained prior to equilibration

ATMStates pydantic-model #

Bases: BaseModel

Configure the lambda schedules.

ATMEquilibrateStage pydantic-model #

Bases: BaseModel

Configure how the system will be equilibrated prior to replica exchange.

ATMSamplingStage pydantic-model #

Bases: HREMD

Configure how the system will be sampled using Hamiltonian replica exchange.

temperature pydantic-field #

temperature: OpenMMQuantity[kelvin] = DEFAULT_TEMPERATURE

The temperature to sample at.

n_warmup_steps pydantic-field #

n_warmup_steps: int = 150000

The number of steps to run each replica for before starting hremd trials. All energies gathered during this period will be discarded.

n_steps_per_cycle pydantic-field #

n_steps_per_cycle: int = 1000

The number of steps to propagate the system by before attempting an exchange.

n_cycles pydantic-field #

n_cycles: int = 2500

The number of cycles of 'propagate the system' -> 'exchange replicas' to run.

max_step_retries pydantic-field #

max_step_retries: int = 5

The maximum number of times to attempt to step if a NaN is encountered before raising an exception

swap_mode pydantic-field #

swap_mode: HREMDSwapModeLiteral | None = value

The mode in which to propose state swaps between replicas. This can either be: 'neighbours', only try and swap adjacent states or ii. 'all', try and swap all states stochastically. If None, no replica exchanges will be attempted.

max_swaps pydantic-field #

max_swaps: int | None = None

The maximum number of swap proposals to make if running in 'all' mode. This variable does nothing when running in 'neighbours' mode.

trajectory_interval pydantic-field #

trajectory_interval: int | None = None

The number of cycles to run before saving the current replica states to DCD trajectory files. If None, no trajectories will be saved.

trajectory_enforce_pbc pydantic-field #

trajectory_enforce_pbc: bool = False

Whether to apply periodic boundary conditions when retrieving coordinates for writing to trajectory files.

checkpoint_interval pydantic-field #

checkpoint_interval: int | None = None

The number of cycles to run before saving the current replica states to checkpoint files. If None, no checkpoints will be saved.

ATMConfig pydantic-model #

Bases: BaseModel

Configuration the stages of the ATM calculation.

ATMEdge pydantic-model #

Bases: Edge

Defines an ATM specific edge in a free energy network.

ligand_1 pydantic-field #

ligand_1: str

The name of the first ligand.

ligand_2 pydantic-field #

ligand_2: str | None

The name of the second ligand. This should be None if running an ABFE calculation.

ligand_1_metadata property #

ligand_1_metadata: dict[str, Any]

Any additional metadata about ligand 1.

ligand_2_metadata property #

ligand_2_metadata: dict[str, Any]

Any additional metadata about ligand 2.

ATMNetwork pydantic-model #

Bases: Network

Defines an ATM specific free energy network.

receptor pydantic-field #

receptor: str | None = None

The name of the receptor. If None, the receptor will be identified from the input directory structure

receptor_metadata property #

receptor_metadata: dict[str, Any]

Any additional metadata about the receptor.

load_config #

load_config(path: Path) -> ATMConfig

Load a configuration from a YAML file.

Parameters:

  • path (Path) –

    The path to the YAML configuration.

Returns:

Source code in femto/fe/atm/_config.py 
375
376
377
378
379
380
381
382
383
384
def load_config(path: pathlib.Path) -> ATMConfig:
    """Load a configuration from a YAML file.

    Args:
        path: The path to the YAML configuration.

    Returns:
        The loaded configuration.
    """
    return ATMConfig(**yaml.safe_load(path.read_text()))

create_state_dicts #

create_state_dicts(
    states: ATMSamplingStage,
) -> list[dict[str, float]]

Map the lambda states specified in the configuration to a list of dictionaries.

Parameters:

Returns:

  • list[dict[str, float]] –

    The dictionaries of lambda states.

Source code in femto/fe/atm/_utils.py 
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
def create_state_dicts(
    states: "femto.fe.atm.ATMSamplingStage",
) -> list[dict[str, float]]:
    """Map the lambda states specified in the configuration to a list of dictionaries.

    Args:
        states: The lambda states.

    Returns:
        The dictionaries of lambda states.
    """
    return [
        {
            openmm.ATMForce.Lambda1(): states.lambda_1[i],
            openmm.ATMForce.Lambda2(): states.lambda_2[i],
            openmm.ATMForce.Direction(): states.direction[i],
            openmm.ATMForce.Alpha(): (
                states.alpha[i] * states.alpha_unit
            ).value_in_unit(openmm.unit.kilojoules_per_mole**-1),
            openmm.ATMForce.Uh(): (states.u0[i] * states.u0_unit).value_in_unit(
                openmm.unit.kilojoules_per_mole
            ),
            openmm.ATMForce.W0(): (states.w0[i] * states.w0_unit).value_in_unit(
                openmm.unit.kilojoules_per_mole
            ),
            **(
                {femto.md.rest.REST_CTX_PARAM: states.bm_b0[i]}
                if states.bm_b0 is not None
                else {}
            ),
        }
        for i in range(len(states.lambda_1))
    ]

compute_ddg #

compute_ddg(
    config: ATMSamplingStage,
    states: ATMStates,
    u_kn: ndarray,
    n_k: ndarray,
) -> DataFrame

Computes the total binding free energy.

Parameters:

  • config (ATMSamplingStage) –

    The sampling configuration.

  • states (ATMStates) –

    The sampled states.

  • u_kn (ndarray) –

    The samples.

  • n_k (ndarray) –

    The sample counts.

Returns:

  • DataFrame –

    A pandas DataFrame containing the total binding free energy and its components.

Source code in femto/fe/atm/_analyze.py 
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
def compute_ddg(
    config: "femto.fe.atm.ATMSamplingStage",
    states: "femto.fe.atm.ATMStates",
    u_kn: numpy.ndarray,
    n_k: numpy.ndarray,
) -> "pandas.DataFrame":
    """Computes the total binding free energy.

    Args:
        config: The sampling configuration.
        states: The sampled states.
        u_kn: The samples.
        n_k: The sample counts.

    Returns:
        A pandas DataFrame containing the total binding free energy and its components.
    """
    import pandas

    n_states = len(states.lambda_1)
    n_states_leg_1 = states.direction.index(-1)

    state_groups = [(n_states_leg_1, 1.0), (n_states - n_states_leg_1, 1.0)]

    estimated, _ = femto.fe.ddg.estimate_ddg(
        u_kn, n_k, config.temperature, state_groups
    )
    return pandas.DataFrame([estimated])

equilibrate_states #

equilibrate_states(
    system: System,
    topology: Structure,
    states: ATMStates,
    config: ATMEquilibrateStage,
    offset: Quantity,
    platform: OpenMMPlatform,
    reporter: Reporter | None = None,
) -> list[State]

Equilibrate the system at each lambda window.

Parameters:

  • system (System) –

    The system to simulate.

  • topology (Structure) –

    The topology of the system to simulate.

  • states (ATMStates) –

    The states of the system to simulate.

  • config (ATMEquilibrateStage) –

    Configuration settings.

  • offset (Quantity) –

    The vector to offset the ligand by using the ATM force.

  • platform (OpenMMPlatform) –

    The accelerator to use.

  • reporter (Reporter | None, default: None ) –

    The (optional) reporter to use to record system statistics such as volume and energy.

Returns:

  • list[State] –

    The final equilibrated state.

Source code in femto/fe/atm/_equilibrate.py 
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
def equilibrate_states(
    system: openmm.System,
    topology: parmed.Structure,
    states: "femto.fe.atm.ATMStates",
    config: "femto.fe.atm.ATMEquilibrateStage",
    offset: openmm.unit.Quantity,
    platform: femto.md.constants.OpenMMPlatform,
    reporter: femto.md.reporting.Reporter | None = None,
) -> list[openmm.State]:
    """Equilibrate the system at each lambda window.

    Args:
        system: The system to simulate.
        topology: The topology of the system to simulate.
        states: The states of the system to simulate.
        config: Configuration settings.
        offset: The vector to offset the ligand by using the ATM force.
        platform: The accelerator to use.
        reporter: The (optional) reporter to use to record system statistics such as
            volume and energy.

    Returns:
        The final equilibrated state.
    """
    import femto.fe.atm._utils

    reporter = femto.md.reporting.NullReporter() if reporter is None else reporter

    openmm_reporter = femto.md.reporting.openmm.OpenMMStateReporter(
        reporter, "equilibration", config.report_interval
    )

    state_dicts = femto.fe.atm._utils.create_state_dicts(states)

    system = copy.deepcopy(system)
    femto.fe.atm._utils.add_atm_force(system, topology, config.soft_core, offset)

    equilibrated_coords = femto.md.simulate.simulate_states(
        system, topology, state_dicts, config.stages, platform, openmm_reporter
    )

    box_vectors = [
        coords.getPeriodicBoxVectors(asNumpy=True).value_in_unit(openmm.unit.angstrom)
        for coords in equilibrated_coords
    ]
    box_vectors_max = numpy.max(numpy.stack(box_vectors), axis=0) * openmm.unit.angstrom

    for i, coords in enumerate(equilibrated_coords):
        context = openmm.Context(system, openmm.VerletIntegrator(0.00001))
        context.setState(coords)
        context.setPeriodicBoxVectors(*box_vectors_max)

        equilibrated_coords[i] = context.getState(getPositions=True)

    return equilibrated_coords

run_hremd #

run_hremd(
    system: System,
    topology: Structure,
    coords: list[State],
    states: ATMStates,
    config: ATMSamplingStage,
    offset: Quantity,
    platform: OpenMMPlatform,
    output_dir: Path,
    reporter: Reporter | None = None,
)

Perform replica exchange sampling for a system prepared for ATM calculations.

Parameters:

  • system (System) –

    The system to simulate. It should not already contain an ATM force.

  • topology (Structure) –

    The topology associated with the system.

  • coords (list[State]) –

    The starting coordinates for each state.

  • states (ATMStates) –

    The lambda states to sample.

  • config (ATMSamplingStage) –

    Configuration settings.

  • offset (Quantity) –

    The vector to offset the ligand by using the ATM force.

  • platform (OpenMMPlatform) –

    The platform to run on.

  • output_dir (Path) –

    The directory to store the sampled energies and statistics to, and any trajectory files if requested.

  • reporter (Reporter | None, default: None ) –

    The reporter to log statistics such as online estimates of the free energy to.

Source code in femto/fe/atm/_sample.py 
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
def run_hremd(
    system: openmm.System,
    topology: parmed.Structure,
    coords: list[openmm.State],
    states: "femto.fe.atm.ATMStates",
    config: "femto.fe.atm.ATMSamplingStage",
    offset: openmm.unit.Quantity,
    platform: femto.md.constants.OpenMMPlatform,
    output_dir: pathlib.Path,
    reporter: femto.md.reporting.Reporter | None = None,
):
    """Perform replica exchange sampling for a system prepared for ATM calculations.

    Args:
        system: The system to simulate. It should *not* already contain an ATM force.
        topology: The topology associated with the system.
        coords: The starting coordinates for each state.
        states: The lambda states to sample.
        config: Configuration settings.
        offset: The vector to offset the ligand by using the ATM force.
        platform: The platform to run on.
        output_dir: The directory to store the sampled energies and statistics to, and
            any trajectory files if requested.
        reporter: The reporter to log statistics such as online estimates of the
            free energy to.
    """
    import femto.fe.atm._utils

    state_dicts = femto.fe.atm._utils.create_state_dicts(states)

    system = copy.deepcopy(system)
    femto.fe.atm._utils.add_atm_force(system, topology, config.soft_core, offset)

    integrator = femto.md.utils.openmm.create_integrator(
        config.integrator, config.temperature
    )
    simulation = femto.md.utils.openmm.create_simulation(
        system, topology, coords[0], integrator, state_dicts[0], platform
    )

    n_states = len(state_dicts)

    swap_mask = {
        (i, j)
        for i in range(n_states)
        for j in range(n_states)
        if states.direction[i] != states.direction[j]
    }

    analysis_fn = None

    if reporter is not None and config.analysis_interval is not None:
        analysis_fn = functools.partial(
            _analyze, config=config, states=states, reporter=reporter
        )

    return femto.md.hremd.run_hremd(
        simulation,
        state_dicts,
        config,
        output_dir,
        swap_mask,
        initial_coords=coords,
        analysis_fn=analysis_fn,
        analysis_interval=config.analysis_interval,
    )

setup_system #

setup_system(
    config: ATMSetupStage,
    receptor: AmberParm,
    ligand_1: AmberParm,
    ligand_2: AmberParm | None,
    displacement: Quantity,
    receptor_ref_query: str | None,
    ligand_1_ref_query: tuple[str, str, str] | None = None,
    ligand_2_ref_query: tuple[str, str, str] | None = None,
) -> tuple[Structure, System]

Prepares a system ready for running the ATM method.

Returns:

  • tuple[Structure, System] –

    The prepared topology and OpenMM system object.

Source code in femto/fe/atm/_setup.py 
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
def setup_system(
    config: "femto.fe.atm.ATMSetupStage",
    receptor: parmed.amber.AmberParm,
    ligand_1: parmed.amber.AmberParm,
    ligand_2: parmed.amber.AmberParm | None,
    displacement: openmm.unit.Quantity,
    receptor_ref_query: str | None,
    ligand_1_ref_query: tuple[str, str, str] | None = None,
    ligand_2_ref_query: tuple[str, str, str] | None = None,
) -> tuple[parmed.Structure, openmm.System]:
    """Prepares a system ready for running the ATM method.

    Returns:
        The prepared topology and OpenMM system object.
    """

    _LOGGER.info(f"setting up an {'ABFE' if ligand_2 is None else 'RBFE'} calculation")

    if receptor_ref_query is None:
        # we need to select the receptor cavity atoms before offsetting any ligands
        # as the query is distance based

        _LOGGER.info("selecting receptor reference atoms")
        receptor_ref_query = femto.fe.reference.select_protein_cavity_atoms(
            receptor,
            [ligand_1] + ([] if ligand_2 is None else [ligand_2]),
            config.reference.receptor_cutoff,
        )

    ligand_1_ref_idxs, ligand_2_ref_idxs = None, None

    # we carve out a 'cavity' where the first ligand will be displaced into during the
    # ATM calculations. this should make equilibration at all states easier.
    cavity_formers = [_offset_ligand(ligand_1, displacement)]

    if ligand_2 is not None:
        # we make sure that when placing solvent molecules we don't accidentally place
        # any on top of the ligands in the cavity itself
        cavity_formers.append(ligand_2)

        (
            ligand_1_ref_idxs,
            ligand_2_ref_idxs,
        ) = femto.fe.reference.select_ligand_idxs(
            ligand_1,
            ligand_2,
            config.reference.ligand_method,
            ligand_1_ref_query,
            ligand_2_ref_query,
        )
        assert ligand_2_ref_idxs is not None, "ligand 2 ref atoms were not selected"
        ligand_2_ref_idxs = tuple(i + len(ligand_1.atoms) for i in ligand_2_ref_idxs)

        ligand_2 = _offset_ligand(ligand_2, displacement)

    _LOGGER.info("solvating system")
    topology = femto.md.solvate.solvate_system(
        receptor,
        ligand_1,
        ligand_2,
        config.solvent,
        displacement,
        cavity_formers=cavity_formers,
    )

    _LOGGER.info("creating OpenMM system")
    system = topology.createSystem(
        nonbondedMethod=openmm.app.PME,
        nonbondedCutoff=0.9 * openmm.unit.nanometer,
        constraints=openmm.app.HBonds,
        rigidWater=True,
    )

    if config.apply_hmr:
        _LOGGER.info("applying HMR.")

        hydrogen_mass = config.hydrogen_mass
        femto.md.system.apply_hmr(system, topology, hydrogen_mass)

    ligand_1_idxs = list(range(len(ligand_1.atoms)))
    ligand_2_idxs = None

    if ligand_2 is not None:
        ligand_2_idxs = [i + len(ligand_1_idxs) for i in range(len(ligand_2.atoms))]

    if config.apply_rest:
        _LOGGER.info("applying REST2.")

        solute_idxs = ligand_1_idxs + ([] if ligand_2_idxs is None else ligand_2_idxs)
        femto.md.rest.apply_rest(system, set(solute_idxs), config.rest_config)

    _LOGGER.info("applying restraints.")
    ligands = [ligand_1] + ([] if ligand_2 is None else [ligand_2])
    idx_offset = sum(len(ligand.atoms) for ligand in ligands)

    receptor_ref_mask = parmed.amber.AmberMask(receptor, receptor_ref_query).Selection()
    receptor_ref_idxs = [i + idx_offset for i, m in enumerate(receptor_ref_mask) if m]
    _LOGGER.info(f"receptor ref idxs={receptor_ref_idxs}")

    _apply_atm_restraints(
        system,
        config.restraints,
        ligand_1_com_idxs=ligand_1_idxs,
        ligand_1_ref_idxs=ligand_1_ref_idxs,
        ligand_2_com_idxs=ligand_2_idxs,
        ligand_2_ref_idxs=ligand_2_ref_idxs,
        receptor_ref_idxs=receptor_ref_idxs,
        offset=displacement,
    )

    restraint_query = config.restraints.receptor_query
    restraint_mask = parmed.amber.AmberMask(receptor, restraint_query).Selection()
    restraint_idxs = [i + idx_offset for i, match in enumerate(restraint_mask) if match]

    _apply_receptor_restraints(
        system, config.restraints, {i: topology.positions[i] for i in restraint_idxs}
    )
    femto.md.utils.openmm.assign_force_groups(system)

    return topology, system

select_displacement #

select_displacement(
    receptor: AmberParm,
    ligand_1: AmberParm,
    ligand_2: AmberParm | None,
    distance: Quantity,
) -> Quantity

Attempts to automatically select a displacement vector for the ligands.

Parameters:

  • receptor (AmberParm) –

    The receptor.

  • ligand_1 (AmberParm) –

    The first ligand positioned in the binding site.

  • ligand_2 (AmberParm | None) –

    The second ligand positioned in the binding site.

  • distance (Quantity) –

    The distance to translate ligands along the displacement vector by.

Returns:

  • Quantity –

    The displacement vector.

Source code in femto/fe/atm/_setup.py 
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
def select_displacement(
    receptor: parmed.amber.AmberParm,
    ligand_1: parmed.amber.AmberParm,
    ligand_2: parmed.amber.AmberParm | None,
    distance: openmm.unit.Quantity,
) -> openmm.unit.Quantity:
    """Attempts to automatically select a displacement vector for the ligands.

    Args:
        receptor: The receptor.
        ligand_1: The first ligand positioned in the binding site.
        ligand_2: The second ligand positioned in the binding site.
        distance: The distance to translate ligands along the displacement vector by.

    Returns:
        The displacement vector.
    """

    ligand_coords = numpy.vstack(
        [ligand_1.coordinates] + ([] if ligand_2 is None else [ligand_2.coordinates])
    )
    receptor_coords = receptor.coordinates

    directions = numpy.array(
        [
            [-1.0, -1.0, -1.0],
            [+1.0, -1.0, -1.0],
            [+1.0, +1.0, -1.0],
            [-1.0, +1.0, -1.0],
            [-1.0, -1.0, +1.0],
            [+1.0, -1.0, +1.0],
            [+1.0, +1.0, +1.0],
            [-1.0, +1.0, +1.0],
        ]
    )
    directions /= numpy.linalg.norm(directions, axis=1, keepdims=True)

    closest_distances = []

    for direction in directions:
        displacement = direction * distance.value_in_unit(openmm.unit.angstrom)

        offset_coords = ligand_coords + displacement

        distances = scipy.spatial.distance.cdist(offset_coords, receptor_coords)
        closest_distances.append(distances.min())

    direction = directions[numpy.argmax(closest_distances)]
    return direction.flatten() * distance

run_workflow #

run_workflow(
    config: ATMConfig,
    ligand_1_coords: Path,
    ligand_1_params: Path,
    ligand_2_coords: Path | None,
    ligand_2_params: Path | None,
    receptor_coords: Path,
    receptor_params: Path | None,
    output_dir: Path,
    report_dir: Path | None = None,
    displacement: Quantity | None = None,
    ligand_1_ref_atoms: tuple[str, str, str] | None = None,
    ligand_2_ref_atoms: tuple[str, str, str] | None = None,
    receptor_ref_atoms: str | None = None,
)

Run the setup, equilibration, and sampling phases.

Parameters:

  • config (ATMConfig) –

    The configuration.

  • ligand_1_coords (Path) –

    The path to the first ligand coordinates.

  • ligand_1_params (Path) –

    The path to the first ligand parameters.

  • ligand_2_coords (Path | None) –

    The path to the second ligand coordinates.

  • ligand_2_params (Path | None) –

    The path to the second ligand parameters.

  • receptor_coords (Path) –

    The path to the receptor coordinates.

  • receptor_params (Path | None) –

    The path to the receptor parameters.

  • report_dir (Path | None, default: None ) –

    The directory to write any statistics to.

  • output_dir (Path) –

    The directory to store all outputs in.

  • displacement (Quantity | None, default: None ) –

    The displacement to offset the ligands by.

  • ligand_1_ref_atoms (tuple[str, str, str] | None, default: None ) –

    The AMBER style query masks that select the first ligands' reference atoms.

  • ligand_2_ref_atoms (tuple[str, str, str] | None, default: None ) –

    The AMBER style query masks that select the second ligands' reference atoms.

  • receptor_ref_atoms (str | None, default: None ) –

    The AMBER style query mask that selects the receptor atoms that form the binding site.

Source code in femto/fe/atm/_runner.py 
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
def run_workflow(
    config: "femto.fe.atm.ATMConfig",
    ligand_1_coords: pathlib.Path,
    ligand_1_params: pathlib.Path,
    ligand_2_coords: pathlib.Path | None,
    ligand_2_params: pathlib.Path | None,
    receptor_coords: pathlib.Path,
    receptor_params: pathlib.Path | None,
    output_dir: pathlib.Path,
    report_dir: pathlib.Path | None = None,
    displacement: openmm.unit.Quantity | None = None,
    ligand_1_ref_atoms: tuple[str, str, str] | None = None,
    ligand_2_ref_atoms: tuple[str, str, str] | None = None,
    receptor_ref_atoms: str | None = None,
):
    """Run the setup, equilibration, and sampling phases.

    Args:
        config: The configuration.
        ligand_1_coords: The path to the first ligand coordinates.
        ligand_1_params: The path to the first ligand parameters.
        ligand_2_coords: The path to the second ligand coordinates.
        ligand_2_params: The path to the second ligand parameters.
        receptor_coords: The path to the receptor coordinates.
        receptor_params: The path to the receptor parameters.
        report_dir: The directory to write any statistics to.
        output_dir: The directory to store all outputs in.
        displacement: The displacement to offset the ligands by.
        ligand_1_ref_atoms: The AMBER style query masks that select the first ligands'
            reference atoms.
        ligand_2_ref_atoms: The AMBER style query masks that select the second ligands'
            reference atoms.
        receptor_ref_atoms: The AMBER style query mask that selects the receptor atoms
            that form the binding site.
    """
    import femto.fe.atm._equilibrate
    import femto.fe.atm._sample

    reporter = (
        femto.md.reporting.NullReporter()
        if report_dir is None or not femto.md.utils.mpi.is_rank_zero()
        else femto.md.reporting.TensorboardReporter(report_dir)
    )

    topology, system, displacement = _prepare_system(
        config.setup,
        ligand_1_coords,
        ligand_1_params,
        ligand_2_coords,
        ligand_2_params,
        receptor_coords,
        receptor_params,
        displacement,
        ligand_1_ref_atoms,
        ligand_2_ref_atoms,
        receptor_ref_atoms,
        output_dir / "_setup",
    )
    topology.symmetry = None  # needed as attr is lost after pickling by MPI

    equilibrate_dir = output_dir / "_equilibrate"
    equilibrate_dir.mkdir(exist_ok=True, parents=True)

    coord_paths = [
        equilibrate_dir / f"state_{state_idx}.xml"
        for state_idx in range(len(config.states.lambda_1))
    ]

    if any(not path.exists() for path in coord_paths):
        coords = femto.fe.atm._equilibrate.equilibrate_states(
            system,
            topology,
            config.states,
            config.equilibrate,
            displacement,
            femto.md.constants.OpenMMPlatform.CUDA,
            reporter,
        )
        _cache_equilibrate_outputs(coords, coord_paths)
    else:
        coords = [
            openmm.XmlSerializer.deserialize(path.read_text()) for path in coord_paths
        ]

    sample_dir = output_dir / "_sample"
    result_path = output_dir / "ddg.csv"

    if not result_path.exists():
        femto.fe.atm._sample.run_hremd(
            system,
            topology,
            coords,
            config.states,
            config.sample,
            displacement,
            femto.md.constants.OpenMMPlatform.CUDA,
            sample_dir,
            reporter,
        )
        _analyze_results(config, sample_dir, result_path)

submit_network #

submit_network(
    config: ATMConfig,
    network: Network,
    output_dir: Path,
    queue_options: SLURMOptions,
    mpi_command: list[str] | None = None,
) -> list[str]

Submits a set of ATM calculations to the SLURM queueing manager.

Parameters:

  • config (ATMConfig) –

    The configuration.

  • network (Network) –

    The network of edges to run.

  • output_dir (Path) –

    The directory to store any outputs in.

  • queue_options (SLURMOptions) –

    The options to use when submitting the jobs.

  • mpi_command (list[str] | None, default: None ) –

    The mpi runner command to use. The default is "srun --mpi=pmix".

Returns:

  • list[str] –

    The ids of the submitted jobs.

Source code in femto/fe/atm/_runner.py 
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
def submit_network(
    config: "femto.fe.atm.ATMConfig",
    network: femto.fe.inputs.Network,
    output_dir: pathlib.Path,
    queue_options: femto.fe.utils.queue.SLURMOptions,
    mpi_command: list[str] | None = None,
) -> list[str]:
    """Submits a set of ATM calculations to the SLURM queueing manager.

    Args:
        config: The configuration.
        network: The network of edges to run.
        output_dir: The directory to store any outputs in.
        queue_options: The options to use when submitting the jobs.
        mpi_command: The mpi runner command to use. The default is
            ``"srun --mpi=pmix"``.

    Returns:
        The ids of the submitted jobs.
    """

    mpi_command = mpi_command if mpi_command is not None else ["srun", "--mpi=pmix"]

    output_dir.mkdir(exist_ok=True, parents=True)

    date_str = datetime.datetime.now().strftime("%Y-%m-%d--%H-%M-%S")
    config_path = output_dir / f"config-{date_str}.yaml"
    config_path.write_text(config.model_dump_yaml(sort_keys=False))

    femto_command = [
        "femto",
        "atm",
        f"--config={config_path}",
        "run-workflow",
    ]

    slurm_job_ids = []

    for edge in network.edges:
        edge_dir = output_dir / f"{edge.ligand_1.name}~{edge.ligand_2.name}"

        job_id = femto.fe.utils.queue.submit_slurm_job(
            [
                *mpi_command,
                *femto_command,
                *_create_run_flags(network.receptor, "receptor"),
                *_create_run_flags(edge.ligand_1, "ligand-1"),
                *_create_run_flags(edge.ligand_2, "ligand-2"),
                f"--output-dir={edge_dir}",
                f"--report-dir={edge_dir}",
            ],
            queue_options,
            edge_dir / f"run-{date_str}.out",
        )

        slurm_job_ids.append(job_id)

    return slurm_job_ids