stjames 0.0.120__py3-none-any.whl → 0.0.122__py3-none-any.whl

stjames/__init__.py

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

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/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/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/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/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/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/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/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/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/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.dist-info → stjames-0.0.122.dist-info}/METADATA

@@ -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.dist-info → stjames-0.0.122.dist-info}/RECORD

@@ -1,9 +1,9 @@
-stjames/__init__.py,sha256=i_vqziEdoBNXhD7Gjb2_1zLkSx0Ko1N8BnCk6No_T6A,522
+stjames/__init__.py,sha256=eM4xm21MXF7RGJcwFEarw0LDz8r8haPhxxgOnCXhsgU,544
 stjames/_deprecated_solvent_settings.py,sha256=gj5j9p3zakIwSTK5_ndqBXJx--IzjZNxZ75z-wipLOo,450
 stjames/atom.py,sha256=RnlAgWsZZJEZnSBWCeoj6NtWht8mZTf4qDWfT_Ot05U,2704
 stjames/base.py,sha256=YhmNZqFdWYD_lCw8rD2fZtt17nh8g7lUFBnGMezB-u8,1837
 stjames/basis_set.py,sha256=wI3M2q9uPf9jhKpAi4E2DrsyKzloDGLRjAlk7krdYgc,949
-stjames/calculation.py,sha256=O2LwwQ_cOLmDOGXTHA9J71YbUZXigUSbvbLA-fSVm3w,915
+stjames/calculation.py,sha256=jm3LvFzJl54yJQtfrqKvodAkZyAvgN3hMWAsWD4HMM8,1333
 stjames/compute_settings.py,sha256=wuYE6W4WqP3oFHwAeEp7XQ0oMz_MNLucWgrsuJ4kEFQ,268
 stjames/constraint.py,sha256=B6oV0rYjmAWr8gpi5f03gRy_uuqjUURVDVwoez5Cfbg,2442
 stjames/correction.py,sha256=ZVErCcj4TPyZeKrdvXVjHa0tFynsCaoy96QZUVxWFM8,413
@@ -17,7 +17,7 @@ stjames/pdb.py,sha256=6ayVUUdTufkXs_nfpRQuT1yGvymuyICB4L5Wh55OsaA,26507
 stjames/periodic_cell.py,sha256=eV_mArsY_MPEFSrFEsTC-CyCc6V8ITAXdk7yhjjNI7M,1080
 stjames/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 stjames/scf_settings.py,sha256=ecEVP3aKArNorpbP0mapnbc-EMjjzx4jeVW6q9tvqsw,436
-stjames/settings.py,sha256=qZ3yhuMIQNiLs8Z-uEyVReSg2IQAjnDzQ6euHHqE6xI,6128
+stjames/settings.py,sha256=PGSNeNosl9OYLAZ3TGdWCN_4UwXI4qujJ5M9VWduE2o,6493
 stjames/solvent.py,sha256=flrGlwD4QavmXlIxeqlsiQv1F_PWyqBTKTDH9HjHdRg,1246
 stjames/status.py,sha256=KQHDqWSd4kBLow23YLcfOkFdtqN61RFZI-jf2zANWRY,501
 stjames/task.py,sha256=OLINRqe66o7t8arffilwmggrF_7TH0L79u6DhGruxV8,329
@@ -37,23 +37,24 @@ stjames/data/read_nist_isotopes.py,sha256=y10FNjW43QpC45qib7VHsIghEwT7GG5rsNwHdc
 stjames/data/symbol_element.json,sha256=vl_buFusTqBd-muYQtMLtTDLy2OtBI6KkBeqkaWRQrg,1186
 stjames/optimization/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 stjames/optimization/freezing_string_method.py,sha256=eEQBqbYHgJH9gVRLDIFtGuPcsHHMLAAt1hF3jtq70lo,2285
-stjames/workflows/__init__.py,sha256=GiSXuBCA2tE5eXHnVYJvjtoqX_zvCn8t0wpkvK1HEls,3438
+stjames/workflows/__init__.py,sha256=wzLBi71MgMxlxRXw3N7feazCym6qNRJ8e4pRk3UXPHM,3520
 stjames/workflows/admet.py,sha256=qFUpCFiLW-3gzuEjCMNBJ6DEG_vquJcPAsN4SVZRfdE,1289
-stjames/workflows/basic_calculation.py,sha256=sAgHBcNHE72ZbZPB9vyZShALRC4zOVw6It6cpJlbX2A,911
+stjames/workflows/basic_calculation.py,sha256=wIiq2lFRN4nxN0__X_JbPSeaUeJ4tKUyg6NaB1xqoEY,1079
 stjames/workflows/batch_docking.py,sha256=o-t5FkfLlbe60jX1_ZeqWSJJ5vsjEIbTa5GLN3ny_tk,1590
 stjames/workflows/bde.py,sha256=g_In-caftXiimrhfdptHjpfrYQUs3vF58qYmRnaTN8g,10825
 stjames/workflows/conformer.py,sha256=18aO6ngMBeGAmQkBdLGCCHr398RIYr1v2hD2IT1u4cc,3005
-stjames/workflows/conformer_search.py,sha256=1kBUT0yCcTPTCtxg1tlTKHRRXkfNYNqzla_89lDEL9k,15696
+stjames/workflows/conformer_search.py,sha256=b3xbyFdulc8VLtk12a_v_zWaS_bkyi6cVVqNyTucD38,16544
 stjames/workflows/descriptors.py,sha256=T4tc7xdtBdxESGO86KR323jPQ2pgwxBqgV0khA6MEgQ,584
 stjames/workflows/docking.py,sha256=t30kqeFXQ0yrlqvN6Jdwt0SdfnJLDsfK-7yFi0gwNbY,4753
 stjames/workflows/double_ended_ts_search.py,sha256=ovJgEVFc6c3mijCE3TKAY70YvqNmAZ5Y4XgV4-tIxBI,3127
 stjames/workflows/electronic_properties.py,sha256=GT3-NC7w-dbcOJ-3AzJ7LgzH6frTbiH2Iyb9BCa-SvY,4112
-stjames/workflows/fukui.py,sha256=095GDGSSEc5PDD1aoKM8J7icgR5tfwS5Bs9XxFQHge4,2387
+stjames/workflows/fukui.py,sha256=E9dfPnZzgwDR3J_8DdTTSwHcMPelyZxlLCmp_WA8UJo,2424
 stjames/workflows/hydrogen_bond_basicity.py,sha256=q9eXty68ZyCmrB6G_8bfeOT8Ui_IQquRPu6z-3rNreQ,1589
 stjames/workflows/ion_mobility.py,sha256=5vUjEYCnF9sN3dTqqEgWAq0jAfdoFvkLubemoXE8DLc,3125
 stjames/workflows/irc.py,sha256=ZP7icylW8rgo_Uh7h3bmyumn0ru1IyF-61nP5Jnmq3M,3402
 stjames/workflows/macropka.py,sha256=Krj0xXuB-u57Kqlf4bbRiHDUWCpliFr6YPiYqPmYaWk,3803
-stjames/workflows/molecular_dynamics.py,sha256=HqWNxxPSAphfI0DdbTERFkq8UeBjEvhnA_ETv0xw_RY,3522
+stjames/workflows/molecular_dynamics.py,sha256=cgjede9TWf-eXRFeUcM59cyVQAhSduL6L0J0oMrX3xc,3543
+stjames/workflows/msa.py,sha256=nK3KmPQv9Sr7yRjNfDWSXIgBrk3PnhvDYHnsVoqoS-M,773
 stjames/workflows/multistage_opt.py,sha256=UN-4WLsT2WEjO5KqDPrcCkb708Co-ZScHx3g2bto768,16597
 stjames/workflows/nmr.py,sha256=1QEF4SB6dWIr-jzLEZ7V972UnRUOTufOJSHwIGyV3dM,2681
 stjames/workflows/pka.py,sha256=i-jzl2lN0yRWc0tgrWSBCplITEByfRyEQrlUhjnzcBc,4580
@@ -61,14 +62,14 @@ stjames/workflows/pose_analysis_md.py,sha256=dpWVKC-8fPdw6ExIXk9xbeVBDUMUYQECpix
 stjames/workflows/protein_binder_design.py,sha256=KnPKQJTMrSO5xfw64Bh7T0wHck1NtysVwkVgHs1cGws,12013
 stjames/workflows/protein_cofolding.py,sha256=w7Sg_ttU4bcJb7wlVcI_AAsLM9WVAJcU5ucbNb5Iyzw,4326
 stjames/workflows/redox_potential.py,sha256=7S18t9Y3eynSnA3lZbRlvLfdbgeBopdiigLzt1zxg5c,3871
-stjames/workflows/scan.py,sha256=DXQBpa2t2PowAtOwmdgpxaSLq--fEShljzAGSb8Nf5U,2993
+stjames/workflows/scan.py,sha256=lgpvrrFG03GWQj2tWeBqPQVYSCGLgYdnQsU9bAUZQok,3317
 stjames/workflows/solubility.py,sha256=lfCVvJjqEaddLUpK6WBxjB7u12Sci-K95A5_qIMkIRM,3028
 stjames/workflows/spin_states.py,sha256=0degmE-frovgoXweshZyjfjqL7nkbaFoO9YoJhvQnaI,4748
 stjames/workflows/strain.py,sha256=paYxDDQTB1eYP_c2kLVz1-QX7Vpw0LLb3ujnFin_SOM,1834
 stjames/workflows/tautomer.py,sha256=7eYKziGPg8Km6lfowTzSkgJfJ4SHUPrAmnTf8Bi-SB0,1164
-stjames/workflows/workflow.py,sha256=cYNP_tK1afJTscXJzYv5VGe8n2Est6PpthS268oRi1U,2494
-stjames-0.0.120.dist-info/licenses/LICENSE,sha256=i05z7xEhyrg6f8j0lR3XYjShnF-MJGFQ-DnpsZ8yiVI,1084
-stjames-0.0.120.dist-info/METADATA,sha256=G8ujY0SFUr28ySoupBKoRViks6v84Q5bepOSRES9EMc,1725
-stjames-0.0.120.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-stjames-0.0.120.dist-info/top_level.txt,sha256=FYCwxl6quhYOAgG-mnPQcCK8vsVM7B8rIUrO-WrQ_PI,8
-stjames-0.0.120.dist-info/RECORD,,
+stjames/workflows/workflow.py,sha256=nwSX-KO2J9Lgd_uNuoYxS1Lv8rZB-jmTfphb8KJODC8,2720
+stjames-0.0.122.dist-info/licenses/LICENSE,sha256=i05z7xEhyrg6f8j0lR3XYjShnF-MJGFQ-DnpsZ8yiVI,1084
+stjames-0.0.122.dist-info/METADATA,sha256=9McsAFJMWpw8ymFCC3VTRWKU6cXslDyZKGvTHt1u7Og,1725
+stjames-0.0.122.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+stjames-0.0.122.dist-info/top_level.txt,sha256=FYCwxl6quhYOAgG-mnPQcCK8vsVM7B8rIUrO-WrQ_PI,8
+stjames-0.0.122.dist-info/RECORD,,