stjames 0.0.117__py3-none-any.whl → 0.0.119__py3-none-any.whl

stjames/workflows/batch_docking.py

@@ -0,0 +1,46 @@
+"""High-throughput docking workflow."""
+
+from typing import Annotated
+
+from pydantic import AfterValidator, ConfigDict, field_validator
+
+from ..pdb import PDB
+from ..types import UUID, Vector3D, round_list
+from .docking import VinaSettings
+from .workflow import BatchSMILESWorkflow
+
+
+class BatchDockingWorkflow(BatchSMILESWorkflow):
+    """
+    Docking workflow.
+
+    Note that the protein can be supplied either by UUID or raw PDB object.
+    We anticipate that the former will dominate deployed usage, but the latter is handy for isolated testing.
+    If, for whatever reason, the workflow is initialized with both a `target_uuid` and a `target`, the UUID will be ignored.
+
+    Inherited:
+    :param smiles_list: list of SMILES
+
+    New:
+    :param target: PDB of the protein, or the UUID of the protein.
+    :param pocket: center (x, y, z) and size (x, y, z) of the pocket
+    :param docking_settings: how to run each docking calculation
+
+    Results:
+    :param best_scores: the best score for each SMILES string
+    """
+
+    model_config = ConfigDict(arbitrary_types_allowed=True)
+
+    target: PDB | UUID
+    pocket: tuple[Vector3D, Vector3D]
+
+    docking_settings: VinaSettings = VinaSettings()
+    best_scores: Annotated[list[float | None], AfterValidator(round_list(3))] = []
+
+    @field_validator("pocket", mode="after")
+    def validate_pocket(cls, pocket: tuple[Vector3D, Vector3D]) -> tuple[Vector3D, Vector3D]:
+        _center, size = pocket
+        if any(q <= 0 for q in size):
+            raise ValueError(f"Pocket size must be positive, got: {size}")
+        return pocket

stjames/workflows/workflow.py

@@ -22,21 +22,31 @@ class Workflow(Base):
         return repr(self)
 
 
+class ProteinSequence(Base):
+    """
+    Protein sequence metadata including cyclic flag.
+
+    :param sequence: amino-acid sequence string
+    :param cyclic: whether this sequence forms a cyclic peptide (defaults to False)
+    """
+
+    sequence: str
+    cyclic: bool = False
+
+
 class FASTAWorkflow(Workflow):
     """
     Base class for Workflows that operate on protein sequences and SMILES.
 
-    :param initial_protein_sequences: protein sequences of interest
+    :param initial_protein_sequences: proteins to evaluate, either plain sequence strings or ProteinSequence objects with cyclic flags
     :param initial_smiles_list: SMILES strings of interest
+    :param ligand_binding_affinity_index: optional index selecting which ligand affinity to evaluate
     """
 
-    initial_protein_sequences: list[str]
+    initial_protein_sequences: list[ProteinSequence] | list[str]
     initial_smiles_list: list[str] | None = None
     ligand_binding_affinity_index: int | None = None
 
-    def __repr__(self) -> str:
-        return f"<{type(self).__name__} {self.initial_protein_sequences} {self.initial_smiles_list}>"
-
 
 class SMILESWorkflow(Workflow):
     """
@@ -51,6 +61,16 @@ class SMILESWorkflow(Workflow):
         return f"<{type(self).__name__} {self.initial_smiles}>"
 
 
+class BatchSMILESWorkflow(Workflow):
+    """
+    Base class for Workflows that operate on a list of SMILES strings.
+
+    :param smiles_list: SMILES strings of interest
+    """
+
+    smiles_list: list[str]
+
+
 class MoleculeWorkflow(Workflow):
     """
     Base class for Workflows that operate on a single molecule.

{stjames-0.0.117.dist-info → stjames-0.0.119.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: stjames
-Version: 0.0.117
+Version: 0.0.119
 Summary: standardized JSON atom/molecule encoding scheme
 Author-email: Corin Wagen <corin@rowansci.com>
 Project-URL: Homepage, https://github.com/rowansci/stjames

{stjames-0.0.117.dist-info → stjames-0.0.119.dist-info}/RECORD

@@ -40,6 +40,7 @@ stjames/optimization/freezing_string_method.py,sha256=eEQBqbYHgJH9gVRLDIFtGuPcsH
 stjames/workflows/__init__.py,sha256=RoqcRk27PTu9UfHFQQtu5quiBipsx4NGvDPst5S1EOc,3317
 stjames/workflows/admet.py,sha256=qFUpCFiLW-3gzuEjCMNBJ6DEG_vquJcPAsN4SVZRfdE,1289
 stjames/workflows/basic_calculation.py,sha256=sAgHBcNHE72ZbZPB9vyZShALRC4zOVw6It6cpJlbX2A,911
+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
@@ -65,9 +66,9 @@ stjames/workflows/solubility.py,sha256=lfCVvJjqEaddLUpK6WBxjB7u12Sci-K95A5_qIMkI
 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=OE05pt2ZOd8TzTOlBngXCVg9wv_553ZR60VNRPlq0f8,1953
-stjames-0.0.117.dist-info/licenses/LICENSE,sha256=i05z7xEhyrg6f8j0lR3XYjShnF-MJGFQ-DnpsZ8yiVI,1084
-stjames-0.0.117.dist-info/METADATA,sha256=LH_VdqC55gK6JJ0beT4uW6xifSRAfMJLQoIh3VpMAHg,1725
-stjames-0.0.117.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-stjames-0.0.117.dist-info/top_level.txt,sha256=FYCwxl6quhYOAgG-mnPQcCK8vsVM7B8rIUrO-WrQ_PI,8
-stjames-0.0.117.dist-info/RECORD,,
+stjames/workflows/workflow.py,sha256=cYNP_tK1afJTscXJzYv5VGe8n2Est6PpthS268oRi1U,2494
+stjames-0.0.119.dist-info/licenses/LICENSE,sha256=i05z7xEhyrg6f8j0lR3XYjShnF-MJGFQ-DnpsZ8yiVI,1084
+stjames-0.0.119.dist-info/METADATA,sha256=zSXQ6WwjJ7Fv7dQ1y_4Nx5Glq6zYzXkq94uKLLDEAbg,1725
+stjames-0.0.119.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+stjames-0.0.119.dist-info/top_level.txt,sha256=FYCwxl6quhYOAgG-mnPQcCK8vsVM7B8rIUrO-WrQ_PI,8
+stjames-0.0.119.dist-info/RECORD,,