Merge pull request #84 from EnzymeML/retain-unmodeled-species

JR-1991 · web-flow · commit d7ec01d0daee · 2025-09-17T12:05:03.000+02:00
Retain unmodeled species
diff --git a/pyenzyme/thinlayers/base.py b/pyenzyme/thinlayers/base.py
@@ -30,23 +30,107 @@ class BaseThinLayer(ABC):
 
     enzmldoc: v2.EnzymeMLDocument
     measurement_ids: List[str]
+    exclude_unmodeled_species: bool = True
 
     def __init__(
         self,
         enzmldoc: v2.EnzymeMLDocument,
         measurement_ids: Optional[List[str]] = None,
         df_per_measurement: bool = False,
+        exclude_unmodeled_species: bool = True,
     ):
         assert isinstance(enzmldoc, v2.EnzymeMLDocument)
         assert isinstance(measurement_ids, list) or measurement_ids is None
 
+        # Remove empty measurements
+        enzmldoc.measurements = [
+            meas for meas in enzmldoc.measurements if meas.species_data
+        ]
+
         if measurement_ids is None:
             measurement_ids = [meas.id for meas in enzmldoc.measurements]
 
-        self.enzmldoc = enzmldoc
+        self.enzmldoc = enzmldoc.model_copy(deep=True)
         self.fitted_doc = enzmldoc.model_copy(deep=True)
         self.measurement_ids = measurement_ids
         self.df_per_measurement = df_per_measurement
+        self.exclude_unmodeled_species = exclude_unmodeled_species
+
+    @staticmethod
+    def _remove_unmodeled_species(enzmldoc: v2.EnzymeMLDocument) -> v2.EnzymeMLDocument:
+        """
+        Removes species that are not modeled from the EnzymeML document.
+
+        This method filters out species that are not referenced in any reactions or ODEs,
+        cleaning up the document to only include modeled species. It also removes
+        measurements that have no remaining species data after filtering.
+
+        Args:
+            enzmldoc (v2.EnzymeMLDocument): The EnzymeML document to filter.
+
+        Returns:
+            v2.EnzymeMLDocument: A deep copy of the document with unmodeled species removed.
+
+        Note:
+            - Creates a deep copy to avoid modifying the original document
+            - Removes measurements that become empty after species filtering
+            - Only considers species from reactions (reactants/products) and ODE equations
+        """
+        enzmldoc = enzmldoc.model_copy(deep=True)
+
+        # Collect all species that are explicitly modeled
+        modeled_species = set()
+
+        # Add species from reactions (reactants and products)
+        for reaction in enzmldoc.reactions:
+            modeled_species.update(
+                reactant.species_id for reactant in reaction.reactants
+            )
+            modeled_species.update(product.species_id for product in reaction.products)
+
+        # Add species from ODE equations
+        modeled_species.update(
+            equation.species_id
+            for equation in enzmldoc.equations
+            if equation.equation_type == v2.EquationType.ODE
+        )
+
+        if not modeled_species:
+            enzmldoc.measurements = []
+            enzmldoc.small_molecules = []
+            enzmldoc.proteins = []
+            enzmldoc.complexes = []
+            return enzmldoc
+
+        filtered_measurements = []
+        for measurement in enzmldoc.measurements:
+            # Filter species data to only include modeled species
+            filtered_species_data = [
+                data
+                for data in measurement.species_data
+                if data.species_id in modeled_species
+            ]
+
+            # Only keep measurements that still have species data
+            if filtered_species_data:
+                measurement.species_data = filtered_species_data
+                filtered_measurements.append(measurement)
+
+        # Update all collections to only include modeled species
+        enzmldoc.measurements = filtered_measurements
+        enzmldoc.small_molecules = [
+            species
+            for species in enzmldoc.small_molecules
+            if species.id in modeled_species
+        ]
+        enzmldoc.proteins = [
+            protein for protein in enzmldoc.proteins if protein.id in modeled_species
+        ]
+        enzmldoc.complexes = [
+            complex for complex in enzmldoc.complexes if complex.id in modeled_species
+        ]
+
+        return enzmldoc
 
     @abstractmethod
     def integrate(
@@ -177,7 +261,12 @@ def df(self) -> pd.DataFrame:
         Raises:
             ValueError: If the conversion doesn't return a DataFrame.
         """
-        df = pe.to_pandas(self.enzmldoc, per_measurement=False)
+        if self.exclude_unmodeled_species:
+            enzmldoc = self._remove_unmodeled_species(self.enzmldoc)
+        else:
+            enzmldoc = self.enzmldoc
+
+        df = pe.to_pandas(enzmldoc, per_measurement=False)
 
         # Drop all this rows where "id" is within measurement_ids
         df = (
diff --git a/pyenzyme/thinlayers/psyces.py b/pyenzyme/thinlayers/psyces.py
@@ -6,10 +6,10 @@
 
 from __future__ import annotations
 
-from copy import deepcopy
 from dataclasses import dataclass
 from pathlib import Path
 from joblib import Parallel, delayed
+import dill
 import numpy as np
 import pandas as pd
 import os
@@ -89,6 +89,7 @@ def __init__(
             enzmldoc=enzmldoc,
             measurement_ids=measurement_ids,
             df_per_measurement=False,
+            exclude_unmodeled_species=True,
         )
 
         if not isinstance(model_dir, Path):
@@ -309,9 +310,10 @@ def _get_experimental_data(self):
 
         Populates the inits, experimental_data, and cols attributes.
         """
+        enzmldoc = self._remove_unmodeled_species(self.enzmldoc)
         self.inits = [
             InitMap.from_measurement(measurement, self.df_map[measurement.id])
-            for measurement in self.enzmldoc.measurements
+            for measurement in enzmldoc.measurements
             if measurement.id in self.measurement_ids
         ]
 
@@ -526,7 +528,7 @@ def to_pysces_model(self, model: pysces.model):
         Returns:
             pysces.model: The updated model with initial conditions set.
         """
-        model = deepcopy(model)
+        model = dill.loads(dill.dumps(model))
         model.sim_time = np.array(self.time)
         model.__dict__.update(
             {
diff --git a/pyproject.toml b/pyproject.toml
@@ -23,6 +23,7 @@ mdmodels = "^0.2.1"
 joblib = "^1.5.0"
 bokeh = "^3.7.3"
 matplotlib = "^3.10"
+dill = ">=0.3.9,<0.5.0"
 
 [tool.poetry.group.psyces.dependencies]
 pysces = "^1.2.3"
diff --git a/tests/unit/test_thinlayer.py b/tests/unit/test_thinlayer.py
@@ -0,0 +1,161 @@
+from pyenzyme.thinlayers.base import BaseThinLayer
+from pyenzyme.versions.v2 import EnzymeMLDocument, EquationType
+
+# Mock data for creating test species measurements
+MOCK_DATA = {
+    "initial": 1.0,
+    "time": [1.0, 2.0, 3.0, 4.0],
+    "data": [1.0, 2.0, 3.0, 4.0],
+}
+
+
+class TestThinLayer:
+    """Test suite for BaseThinLayer functionality."""
+
+    def test_remove_unmodeled_species_reaction(self):
+        """
+        Test that unmodeled species are removed when they're not part of any reaction.
+
+        This test verifies that:
+        - Species not referenced in reactions are removed from the document
+        - Measurements containing only unmodeled species are removed
+        - Measurements with mixed modeled/unmodeled species keep only modeled ones
+        """
+        enzmldoc = self._create_enzmldoc()
+
+        # Add reaction with only Substrate and Product (Unmodeled is not included)
+        reaction = enzmldoc.add_to_reactions(id="R1", name="R1")
+        reaction.add_to_reactants(species_id="Substrate", stoichiometry=1)
+        reaction.add_to_products(species_id="Product", stoichiometry=1)
+
+        # Remove unmodeled species
+        thinlayer = MockThinLayer(enzmldoc)
+        tl_enzmldoc = thinlayer.optimize()
+
+        assert len(tl_enzmldoc.small_molecules) == 2, (
+            f"Unmodeled small molecules should be removed, but {len(tl_enzmldoc.small_molecules)} remain."
+        )
+        assert len(tl_enzmldoc.measurements) == 2, (
+            f"Unmodeled measurements should be removed, but {len(tl_enzmldoc.measurements)} remain."
+        )
+
+        measurement_has_unmodeled: list[str] = []
+
+        for measurement in tl_enzmldoc.measurements:
+            for species_data in measurement.species_data:
+                if species_data.species_id == "Unmodeled":
+                    measurement_has_unmodeled.append(measurement.id)
+
+        assert len(measurement_has_unmodeled) == 0, (
+            f"Unmodeled species should be removed, but appears in measurements {measurement_has_unmodeled}."
+        )
+
+    def test_remove_unmodeled_species_odes(self):
+        """
+        Test that unmodeled species are removed when they're not part of any ODE.
+
+        This test verifies that:
+        - Species not referenced in ODE equations are removed from the document
+        - Measurements containing only unmodeled species are removed
+        - Measurements with mixed modeled/unmodeled species keep only modeled ones
+        """
+        enzmldoc = self._create_enzmldoc()
+
+        # Add ODEs with only Substrate and Product (Unmodeled is not included)
+        enzmldoc.add_to_equations(
+            species_id="Substrate",
+            equation_type=EquationType.ODE,
+            equation="-Substrate",
+        )
+
+        enzmldoc.add_to_equations(
+            species_id="Product",
+            equation_type=EquationType.ODE,
+            equation="Substrate",
+        )
+
+        # Remove unmodeled species
+        thinlayer = MockThinLayer(enzmldoc)
+        tl_enzmldoc = thinlayer.optimize()
+
+        assert len(tl_enzmldoc.small_molecules) == 2, (
+            f"Unmodeled small molecules should be removed, but {len(tl_enzmldoc.small_molecules)} remain."
+        )
+        assert len(tl_enzmldoc.measurements) == 2, (
+            f"Unmodeled measurements should be removed, but {len(tl_enzmldoc.measurements)} remain."
+        )
+
+        measurement_has_unmodeled: list[str] = []
+
+        for measurement in tl_enzmldoc.measurements:
+            for species_data in measurement.species_data:
+                if species_data.species_id == "Unmodeled":
+                    measurement_has_unmodeled.append(measurement.id)
+
+        assert len(measurement_has_unmodeled) == 0, (
+            f"Unmodeled species should be removed, but appears in measurements {measurement_has_unmodeled}."
+        )
+
+    def _create_enzmldoc(self) -> EnzymeMLDocument:
+        """
+        Create a test EnzymeML document with various measurement scenarios.
+
+        Creates a document with:
+        - Three species: Substrate, Product, and Unmodeled
+        - Four measurements:
+          - M1: Contains all three species (mixed modeled/unmodeled)
+          - M2: Contains only modeled species (Substrate, Product)
+          - M3: Contains only unmodeled species (Unmodeled)
+          - M4: Empty measurement (no species data)
+
+        Returns:
+            EnzymeMLDocument: A test document for use in unit tests.
+        """
+        enzmldoc = EnzymeMLDocument(name="Test")
+
+        # Add small molecules
+        substrate = enzmldoc.add_to_small_molecules(id="Substrate", name="Substrate")
+        product = enzmldoc.add_to_small_molecules(id="Product", name="Product")
+        unmodeled = enzmldoc.add_to_small_molecules(id="Unmodeled", name="Unmodeled")
+
+        # Add a measurement with unmodeled species
+        measurement = enzmldoc.add_to_measurements(id="M1", name="M1")
+        measurement.add_to_species_data(species_id=substrate.id, **MOCK_DATA)
+        measurement.add_to_species_data(species_id=product.id, **MOCK_DATA)
+        measurement.add_to_species_data(species_id=unmodeled.id, **MOCK_DATA)
+
+        # Add a Measurement only with modeled species
+        measurement = enzmldoc.add_to_measurements(id="M2", name="M2")
+        measurement.add_to_species_data(species_id=substrate.id, **MOCK_DATA)
+        measurement.add_to_species_data(species_id=product.id, **MOCK_DATA)
+
+        # Add a Measurement with only unmodeled species
+        measurement = enzmldoc.add_to_measurements(id="M3", name="M3")
+        measurement.add_to_species_data(species_id=unmodeled.id, **MOCK_DATA)
+
+        # Add an empty measurement
+        measurement = enzmldoc.add_to_measurements(id="M4", name="M4")
+
+        return enzmldoc
+
+
+class MockThinLayer(BaseThinLayer):
+    """
+    Mock implementation of BaseThinLayer for testing purposes.
+
+    This class provides minimal implementations of the abstract methods
+    to allow testing of the base class functionality without requiring
+    a full thin layer implementation.
+    """
+
+    def integrate(self, *args, **kwargs):
+        """Mock integration method that does nothing."""
+        pass
+
+    def optimize(self, *args, **kwargs):
+        """Mock optimization method that does nothing."""
+        return self._remove_unmodeled_species(self.enzmldoc)
+
+    def write(self, *args, **kwargs):
+        """Mock write method that does nothing."""
+        pass
