stjames 0.0.120__tar.gz → 0.0.122__tar.gz

{stjames-0.0.120/stjames.egg-info → stjames-0.0.122}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: stjames
-Version: 0.0.120
+Version: 0.0.122
 Summary: standardized JSON atom/molecule encoding scheme
 Author-email: Corin Wagen <corin@rowansci.com>
 Project-URL: Homepage, https://github.com/rowansci/stjames

{stjames-0.0.120 → stjames-0.0.122}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "stjames"
-version = "0.0.120"
+version = "0.0.122"
 description = "standardized JSON atom/molecule encoding scheme"
 readme = "README.md"
 requires-python = ">=3.11"

{stjames-0.0.120 → stjames-0.0.122}/stjames/__init__.py

@@ -22,3 +22,4 @@ from .status import *
 from .constraint import *
 from .message import *
 from .types import *
+from .engine import *

{stjames-0.0.120 → stjames-0.0.122}/stjames/calculation.py

@@ -1,10 +1,13 @@
-from typing import Optional
+from typing import Optional, Self
 
-from .base import Base, LowercaseStrEnum
+from pydantic import model_validator
+
+from .base import Base, LowercaseStrEnum, UniqueList
 from .message import Message
 from .molecule import Molecule
 from .settings import Settings
 from .status import Status
+from .task import Task
 from .types import UUID
 
 
@@ -20,6 +23,7 @@ class StJamesVersion(LowercaseStrEnum):
 class Calculation(Base):
     molecules: list[Molecule]
 
+    tasks: UniqueList[Task] = []
     settings: Settings = Settings()
 
     status: Status = Status.QUEUED
@@ -29,8 +33,17 @@ class Calculation(Base):
     logfile: Optional[str] = None
     messages: list[Message] = []
 
+    # DEPRECATED - moving into settings
     engine: Optional[str] = "peregrine"
+
     uuids: list[UUID | None] | None = None
 
     # not to be changed by end users, diff. versions will have diff. defaults
     json_format: str = StJamesVersion.V0
+
+    @model_validator(mode="after")
+    def populate_tasks(self) -> Self:
+        """Set the tasks from the settings, so that we don't have to migrate old entries."""
+        if len(self.tasks) == 0:
+            self.tasks = self.settings.tasks
+        return self

{stjames-0.0.120 → stjames-0.0.122}/stjames/settings.py

@@ -6,6 +6,7 @@ from .base import Base, UniqueList
 from .basis_set import BasisSet
 from .compute_settings import ComputeSettings
 from .correction import Correction
+from .engine import Engine
 from .method import CORRECTABLE_NNP_METHODS, METHODS_WITH_CORRECTION, PREPACKAGED_METHODS, Method
 from .mode import Mode
 from .opt_settings import OptimizationSettings
@@ -20,11 +21,13 @@ _T = TypeVar("_T")
 class Settings(Base):
     mode: Mode = Mode.AUTO
 
+    # DEPRECATED - specify tasks only in BasicCalculationWorkflow or Calculation now
+    tasks: UniqueList[Task] = [Task.ENERGY, Task.CHARGE, Task.DIPOLE]
+
     method: Method = Method.HARTREE_FOCK
     basis_set: Optional[BasisSet] = None
-    tasks: UniqueList[Task] = [Task.ENERGY, Task.CHARGE, Task.DIPOLE]
+    engine: Engine = None  # type: ignore [assignment]
     corrections: UniqueList[Correction] = []
-
     solvent_settings: Optional[SolventSettings] = None
 
     # scf/opt settings will be set automatically based on mode, but can be overridden manually
@@ -33,6 +36,12 @@ class Settings(Base):
     thermochem_settings: ThermochemistrySettings = ThermochemistrySettings()
     compute_settings: ComputeSettings = ComputeSettings()
 
+    @model_validator(mode="after")
+    def set_engine(self) -> Self:
+        """Set the calculation engine."""
+        self.engine = self.engine or self.method.default_engine()
+        return self
+
     # mypy has this dead wrong (https://docs.pydantic.dev/2.0/usage/computed_fields/)
     # Python 3.12 narrows the reason for the ignore to prop-decorator
     @computed_field  # type: ignore[misc, prop-decorator, unused-ignore]

{stjames-0.0.120 → stjames-0.0.122}/stjames/workflows/__init__.py

@@ -18,6 +18,7 @@ from .ion_mobility import *
 from .irc import *
 from .macropka import *
 from .molecular_dynamics import *
+from .msa import *
 from .multistage_opt import *
 from .nmr import *
 from .pka import *
@@ -49,6 +50,7 @@ WORKFLOW_NAME = Literal[
     "irc",
     "macropka",
     "molecular_dynamics",
+    "msa",
     "multistage_opt",
     "nmr",
     "pka",
@@ -81,6 +83,7 @@ WORKFLOW_MAPPING: dict[WORKFLOW_NAME, Workflow] = {
     "macropka": MacropKaWorkflow,  # type: ignore [dict-item]
     "molecular_dynamics": MolecularDynamicsWorkflow,  # type: ignore [dict-item]
     "multistage_opt": MultiStageOptWorkflow,  # type: ignore [dict-item]
+    "msa": MSAWorkflow,  # type: ignore [dict-item]
     "nmr": NMRSpectroscopyWorkflow,  # type: ignore [dict-item]
     "pka": pKaWorkflow,  # type: ignore [dict-item]
     "pose_analysis_md": PoseAnalysisMolecularDynamicsWorkflow,  # type: ignore [dict-item]

{stjames-0.0.120 → stjames-0.0.122}/stjames/workflows/basic_calculation.py

@@ -4,8 +4,10 @@ from typing import Self
 
 from pydantic import model_validator
 
+from ..base import UniqueList
 from ..engine import Engine
 from ..settings import Settings
+from ..task import Task
 from ..types import UUID
 from .workflow import MoleculeWorkflow
 
@@ -25,9 +27,12 @@ class BasicCalculationWorkflow(MoleculeWorkflow):
     """
 
     settings: Settings
-    engine: Engine = None  # type: ignore [assignment]
+    tasks: UniqueList[Task] = [Task.ENERGY, Task.CHARGE, Task.DIPOLE]
     calculation_uuid: UUID | None = None
 
+    # DEPRECATED - specify in settings now
+    engine: Engine = None  # type: ignore [assignment]
+
     @model_validator(mode="after")
     def set_engine(self) -> Self:
         """Set the calculation engine."""

{stjames-0.0.120 → stjames-0.0.122}/stjames/workflows/conformer_search.py

@@ -3,9 +3,9 @@
 from abc import ABC
 from typing import Annotated, Literal, Self, Sequence, TypeVar
 
-from pydantic import AfterValidator, BaseModel, Field, field_validator, model_validator
+from pydantic import AfterValidator, BaseModel, Field, PositiveInt, field_validator, model_validator
 
-from ..base import LowercaseStrEnum
+from ..base import Base, LowercaseStrEnum
 from ..constraint import Constraint
 from ..method import Method, XTBMethod
 from ..mode import Mode
@@ -41,6 +41,28 @@ class ScreeningSettings(BaseModel):
     max_confs: int | None = None
 
 
+class ConformerClusteringSettings(Base):
+    """
+    Settings for (optionally) clustering conformers based on three-dimensional properties.
+
+    The properties used for clustering are:
+    - Solvent-accessible surface area
+    - Polar solvent-accessible surface area
+    - Radius of gyration
+    - Plane of best fit
+    - Normalized principal moment ratios 1 and 2
+
+    Rowan uses k-means clustering to identify representative conformers.
+    This loosely follows Wilcken and co-workers (10.1007/s10822-020-00337-7).
+
+    :param num_clusters: the number of clusters to include
+    :param conformers_per_cluster: the number of compounds to pick from each cluster
+    """
+
+    num_clusters: PositiveInt = 5
+    conformers_per_cluster: PositiveInt = 3
+
+
 class ConformerGenSettings(BaseModel):
     """
     Conformer generation settings.
@@ -310,6 +332,8 @@ class ConformerGenMixin(BaseModel):
     nci: bool = False
     max_confs: int | None = None
 
+    conformer_clustering_settings: ConformerClusteringSettings | None = None
+
     @model_validator(mode="after")
     def validate_and_build_conf_gen_settings(self) -> Self:
         """Validate and build the ConformerGenSettings."""

{stjames-0.0.120 → stjames-0.0.122}/stjames/workflows/fukui.py

@@ -36,9 +36,13 @@ class FukuiIndexWorkflow(MoleculeWorkflow):
     """
 
     opt_settings: Settings | None = None
+
+    # DEPRECATED - will be removed
     opt_engine: Engine | None = None
 
     fukui_settings: Settings = Settings(method=Method.GFN1_XTB)
+
+    # DEPRECATED - will be removed
     fukui_engine: Engine = None  # type: ignore [assignment]
 
     optimization: UUID | None = None
@@ -52,7 +56,6 @@ class FukuiIndexWorkflow(MoleculeWorkflow):
     def set_engines(self) -> Self:
         """Set the engines for optimization and Fukui index calculations."""
         if self.opt_settings is not None and self.opt_engine is None:
-            self.opt_engine = self.opt_settings.method.default_engine()
-        self.fukui_engine = self.fukui_engine or self.fukui_settings.method.default_engine()
-
+            self.opt_engine = self.opt_settings.engine
+        self.fukui_engine = self.fukui_engine or self.fukui_settings.engine
         return self

{stjames-0.0.120 → stjames-0.0.122}/stjames/workflows/molecular_dynamics.py

@@ -96,6 +96,8 @@ class MolecularDynamicsWorkflow(MoleculeWorkflow):
 
     settings: MolecularDynamicsSettings
     calc_settings: Settings
+
+    # DEPRECATED - will remove later
     calc_engine: Engine = None  # type: ignore[assignment]
 
     frames: list[Frame] = []
@@ -103,6 +105,6 @@ class MolecularDynamicsWorkflow(MoleculeWorkflow):
     @model_validator(mode="after")
     def validate_calc_engine(self) -> Self:
         """Ensure that the calc_engine is set."""
-        self.calc_engine = self.calc_engine or self.calc_settings.method.default_engine()
+        self.calc_engine = self.calc_engine or self.calc_settings.engine
 
         return self

stjames-0.0.122/stjames/workflows/msa.py

@@ -0,0 +1,31 @@
+from ..base import LowercaseStrEnum
+from .workflow import ProteinSequenceWorkflow
+
+
+class MSAFormat(LowercaseStrEnum):
+    """Format of the MSA."""
+
+    COLABFOLD_DEFAULT = "colabfold_default"
+    AF3_JSON = "af3_json"
+
+
+class MSAWorkflow(ProteinSequenceWorkflow):
+    """
+    Workflow for generating a MSA from protein sequences.
+
+    Inherited:
+    :param initial_protein_sequences: protein sequences of interest
+
+    New:
+    :param format: the format of the MSA return files
+
+    Results:
+    :param a3m_file: A3M file string
+    :param m8_file: M8 file string
+    :param af3_json_file: AF3 JSON file string
+    """
+
+    format: MSAFormat = MSAFormat.COLABFOLD_DEFAULT
+    a3m_file: str | None = None
+    m8_file: str | None = None
+    af3_json_file: str | None = None

{stjames-0.0.120 → stjames-0.0.122}/stjames/workflows/scan.py

@@ -1,12 +1,11 @@
-"""Scan workflow."""
-
-from typing import Annotated
+from typing import Annotated, Self
 
 import numpy as np
 from numpy.typing import NDArray
-from pydantic import AfterValidator, field_validator
+from pydantic import AfterValidator, field_validator, model_validator
 
 from ..base import Base, round_optional_float
+from ..engine import Engine
 from ..molecule import Molecule
 from ..settings import Settings
 from ..types import UUID
@@ -75,7 +74,9 @@ class ScanWorkflow(MoleculeWorkflow):
     scan_settings: ScanSettings | list[ScanSettings]
     scan_settings_2d: ScanSettings | list[ScanSettings] = []
     calc_settings: Settings
-    calc_engine: str
+
+    # DEPRECATED - will be removed in future
+    calc_engine: Engine = None  # type: ignore [assignment]
 
     wavefront_propagation: bool = True
 
@@ -94,3 +95,9 @@ class ScanWorkflow(MoleculeWorkflow):
                 raise ValueError("Concerted scan settings must have same number of steps!")
 
         return val
+
+    @model_validator(mode="after")
+    def set_calc_engine(self) -> Self:
+        """Set the calculation engine."""
+        self.calc_engine = self.calc_engine or self.calc_settings.engine
+        return self

{stjames-0.0.120 → stjames-0.0.122}/stjames/workflows/workflow.py

@@ -95,6 +95,16 @@ class MoleculeWorkflow(Workflow):
         return mode
 
 
+class ProteinSequenceWorkflow(Workflow):
+    """
+    Base class for Workflows that operate on protein sequences.
+
+    :param initial_protein_sequences: protein sequences of interest
+    """
+
+    protein_sequences: list[str]
+
+
 class DBCalculation(Base):
     """Encodes a calculation that's in the database. This isn't terribly useful by itself."""
 

{stjames-0.0.120 → stjames-0.0.122/stjames.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: stjames
-Version: 0.0.120
+Version: 0.0.122
 Summary: standardized JSON atom/molecule encoding scheme
 Author-email: Corin Wagen <corin@rowansci.com>
 Project-URL: Homepage, https://github.com/rowansci/stjames

{stjames-0.0.120 → stjames-0.0.122}/stjames.egg-info/SOURCES.txt

@@ -62,6 +62,7 @@ stjames/workflows/ion_mobility.py
 stjames/workflows/irc.py
 stjames/workflows/macropka.py
 stjames/workflows/molecular_dynamics.py
+stjames/workflows/msa.py
 stjames/workflows/multistage_opt.py
 stjames/workflows/nmr.py
 stjames/workflows/pka.py