changes for fix

aditya0by0 · aditya0by0 · commit 2422518024fa · 2025-04-14T19:23:41.000+02:00
diff --git a/chebai/models/base.py b/chebai/models/base.py
@@ -1,5 +1,5 @@
 import logging
-from typing import Any, Dict, Optional, Union, Iterable
+from typing import Any, Dict, Iterable, Optional, Union
 
 import torch
 from lightning.pytorch.core.module import LightningModule
diff --git a/chebai/preprocessing/collect_all.py b/chebai/preprocessing/collect_all.py
@@ -5,6 +5,7 @@
 import pytorch_lightning as pl
 import torch
 import torch.nn.functional as F
+from data import ClassificationData, JCIClassificationData
 from pytorch_lightning import loggers as pl_loggers
 from pytorch_lightning.callbacks import ModelCheckpoint
 from pytorch_lightning.metrics import F1
@@ -13,8 +14,6 @@
 from torch_geometric import nn as tgnn
 from torch_geometric.data import DataLoader
 
-from data import ClassificationData, JCIClassificationData
-
 logging.getLogger("pysmiles").setLevel(logging.CRITICAL)
 
 
diff --git a/chebai/preprocessing/datasets/__init__.py b/chebai/preprocessing/datasets/__init__.py
@@ -1 +1 @@
-from .base import XYBaseDataModule, _DynamicDataset
+from .base import XYBaseDataModule, _DynamicDataset
diff --git a/configs/data/scope/scope50.yml b/configs/data/scope/scope50.yml
@@ -1,3 +1,3 @@
 class_path: chebai.preprocessing.datasets.scope.scope.SCOPeOver50
 init_args:
-  scope_version: "2.08"
+  scope_version: "2.08"
diff --git a/tests/unit/mock_data/ontology_mock_data.py b/tests/unit/mock_data/ontology_mock_data.py
@@ -114,6 +114,298 @@ def get_data_in_dataframe() -> pd.DataFrame:
         pass
 
 
+class ChebiMockOntology(MockOntologyGraphData):
+    """
+    A mock ontology representing a simplified ChEBI (Chemical Entities of Biological Interest) structure.
+    This class is used for testing purposes and includes nodes and edges representing chemical compounds
+    and their relationships in a graph structure.
+
+    Nodes:
+    - CHEBI:12345 (Compound A)
+    - CHEBI:54321 (Compound B)
+    - CHEBI:67890 (Compound C)
+    - CHEBI:11111 (Compound D)
+    - CHEBI:22222 (Compound E)
+    - CHEBI:99999 (Compound F)
+    - CHEBI:77533 (Compound G, Obsolete node)
+    - CHEBI:77564 (Compound H, Obsolete node)
+    - CHEBI:88888 (Compound I)
+
+    Valid Edges:
+    - CHEBI:54321 -> CHEBI:12345
+    - CHEBI:67890 -> CHEBI:12345
+    - CHEBI:67890 -> CHEBI:88888
+    - CHEBI:11111 -> CHEBI:54321
+    - CHEBI:22222 -> CHEBI:67890
+    - CHEBI:12345 -> CHEBI:99999
+
+    The class also includes methods to retrieve nodes, edges, and transitive closure of the graph.
+
+    Visual Representation Graph with Valid Nodes and Edges:
+
+                                       22222
+                                        /
+                       11111         67890
+                         \\         /  \
+                        54321     /    88888
+                           \\   /
+                           12345
+                             \
+                            99999
+    """
+
+    @staticmethod
+    def get_nodes() -> List[int]:
+        """
+        Get the set of valid node IDs in the mock ontology.
+
+        Returns:
+        - Set[int]: A set of integers representing the valid ChEBI node IDs.
+        """
+        return [11111, 12345, 22222, 54321, 67890, 88888, 99999]
+
+    @staticmethod
+    def get_number_of_nodes() -> int:
+        """
+        Get the number of valid nodes in the mock ontology.
+
+        Returns:
+        - int: The number of valid nodes.
+        """
+        return len(ChebiMockOntology.get_nodes())
+
+    @staticmethod
+    def get_edges() -> Set[Tuple[int, int]]:
+        """
+        Get the set of valid edges in the mock ontology.
+
+        Returns:
+        - Set[Tuple[int, int]]: A set of tuples representing the directed edges
+          between ChEBI nodes.
+        """
+        return {
+            (54321, 12345),
+            (67890, 12345),
+            (67890, 88888),
+            (11111, 54321),
+            (22222, 67890),
+            (12345, 99999),
+        }
+
+    @staticmethod
+    def get_number_of_edges() -> int:
+        """
+        Get the number of valid edges in the mock ontology.
+
+        Returns:
+        - int: The number of valid edges.
+        """
+        return len(ChebiMockOntology.get_edges())
+
+    @staticmethod
+    def get_edges_of_transitive_closure_graph() -> Set[Tuple[int, int]]:
+        """
+        Get the set of edges derived from the transitive closure of the mock ontology graph.
+
+        Returns:
+        - Set[Tuple[int, int]]: A set of tuples representing the directed edges
+          in the transitive closure of the ChEBI graph.
+        """
+        return {
+            (54321, 12345),
+            (54321, 99999),
+            (67890, 12345),
+            (67890, 99999),
+            (67890, 88888),
+            (11111, 54321),
+            (11111, 12345),
+            (11111, 99999),
+            (22222, 67890),
+            (22222, 12345),
+            (22222, 99999),
+            (22222, 88888),
+            (12345, 99999),
+        }
+
+    @staticmethod
+    def get_number_of_transitive_edges() -> int:
+        """
+        Get the number of edges in the transitive closure of the mock ontology graph.
+
+        Returns:
+        - int: The number of edges in the transitive closure graph.
+        """
+        return len(ChebiMockOntology.get_edges_of_transitive_closure_graph())
+
+    @staticmethod
+    def get_obsolete_nodes_ids() -> Set[int]:
+        """
+        Get the set of obsolete node IDs in the mock ontology.
+
+        Returns:
+        - Set[int]: A set of integers representing the obsolete ChEBI node IDs.
+        """
+        return {77533, 77564}
+
+    @staticmethod
+    def get_raw_data() -> str:
+        """
+        Get the raw data representing the mock ontology in OBO format.
+
+        Returns:
+        - str: A string containing the raw OBO data for the mock ChEBI terms.
+        """
+        return """
+        [Term]
+        id: CHEBI:12345
+        name: Compound A
+        subset: 2_STAR
+        property_value: http://purl.obolibrary.org/obo/chebi/formula "C26H35ClN4O6S" xsd:string
+        property_value: http://purl.obolibrary.org/obo/chebi/charge "0" xsd:string
+        property_value: http://purl.obolibrary.org/obo/chebi/monoisotopicmass "566.19658" xsd:string
+        property_value: http://purl.obolibrary.org/obo/chebi/mass "567.099" xsd:string
+        property_value: http://purl.obolibrary.org/obo/chebi/inchikey "ROXPMFGZZQEKHB-IUKKYPGJSA-N" xsd:string
+        property_value: http://purl.obolibrary.org/obo/chebi/smiles "C1=CC=CC=C1" xsd:string
+        property_value: http://purl.obolibrary.org/obo/chebi/inchi "InChI=1S/C26H35ClN4O6S/c1-16(2)28-26(34)30(5)14-23-17(3)13-31(18(4)15-32)25(33)21-7-6-8-22(24(21)37-23)29-38(35,36)20-11-9-19(27)10-12-20/h6-12,16-18,23,29,32H,13-15H2,1-5H3,(H,28,34)/t17-,18-,23+/m0/s1" xsd:string
+        xref: LINCS:LSM-20139
+        is_a: CHEBI:54321
+        is_a: CHEBI:67890
+
+        [Term]
+        id: CHEBI:54321
+        name: Compound B
+        property_value: http://purl.obolibrary.org/obo/chebi/smiles "C1=CC=CC=C1O" xsd:string
+        is_a: CHEBI:11111
+        is_a: CHEBI:77564
+
+        [Term]
+        id: CHEBI:67890
+        name: Compound C
+        property_value: http://purl.obolibrary.org/obo/chebi/smiles "C1=CC=CC=C1N" xsd:string
+        is_a: CHEBI:22222
+
+        [Term]
+        id: CHEBI:11111
+        name: Compound D
+        property_value: http://purl.obolibrary.org/obo/chebi/smiles "C1=CC=CC=C1F" xsd:string
+
+        [Term]
+        id: CHEBI:22222
+        name: Compound E
+        property_value: http://purl.obolibrary.org/obo/chebi/smiles "C1=CC=CC=C1Cl" xsd:string
+
+        [Term]
+        id: CHEBI:99999
+        name: Compound F
+        property_value: http://purl.obolibrary.org/obo/chebi/smiles "C1=CC=CC=C1Br" xsd:string
+        is_a: CHEBI:12345
+
+        [Term]
+        id: CHEBI:77533
+        name: Compound G
+        is_a: CHEBI:99999
+        property_value: http://purl.obolibrary.org/obo/chebi/smiles "C1=C1Br" xsd:string
+        is_obsolete: true
+
+        [Term]
+        id: CHEBI:77564
+        name: Compound H
+        property_value: http://purl.obolibrary.org/obo/chebi/smiles "CC=C1Br" xsd:string
+        is_obsolete: true
+
+        [Typedef]
+        id: has_major_microspecies_at_pH_7_3
+        name: has major microspecies at pH 7.3
+        is_cyclic: true
+        is_transitive: false
+
+        [Term]
+        id: CHEBI:88888
+        name: Compound I
+        property_value: http://purl.obolibrary.org/obo/chebi/smiles "C1=CC=CC=C1[Mg+]" xsd:string
+        is_a: CHEBI:67890
+        """
+
+    @staticmethod
+    def get_data_in_dataframe() -> pd.DataFrame:
+        data = OrderedDict(
+            id=[
+                12345,
+                54321,
+                67890,
+                11111,
+                22222,
+                99999,
+                88888,
+            ],
+            name=[
+                "Compound A",
+                "Compound B",
+                "Compound C",
+                "Compound D",
+                "Compound E",
+                "Compound F",
+                "Compound I",
+            ],
+            SMILES=[
+                "C1=CC=CC=C1",
+                "C1=CC=CC=C1O",
+                "C1=CC=CC=C1N",
+                "C1=CC=CC=C1F",
+                "C1=CC=CC=C1Cl",
+                "C1=CC=CC=C1Br",
+                "C1=CC=CC=C1[Mg+]",
+            ],
+            **{
+                # -row- [12345, 54321, 67890, 11111, 22222, 99999, 88888]
+                11111: [True, True, False, True, False, True, False],
+                12345: [True, False, False, False, False, True, False],
+                22222: [True, False, True, False, True, True, True],
+                54321: [True, True, False, False, False, True, False],
+                67890: [True, False, True, False, False, True, True],
+                88888: [False, False, False, False, False, False, True],
+                99999: [False, False, False, False, False, True, False],
+            },
+        )
+
+        data_df = pd.DataFrame(data)
+
+        # ------------- Code Approach -------
+        # ancestors_of_nodes = {}
+        # for parent, child in ChebiMockOntology.get_edges_of_transitive_closure_graph():
+        #     if child not in ancestors_of_nodes:
+        #         ancestors_of_nodes[child] = set()
+        #     if parent not in ancestors_of_nodes:
+        #         ancestors_of_nodes[parent] = set()
+        #     ancestors_of_nodes[child].add(parent)
+        #     ancestors_of_nodes[child].add(child)
+        #
+        # # For each node in the ontology, create a column to check if it's an ancestor of any other node or itself
+        # for node in ChebiMockOntology.get_nodes():
+        #     data_df[node] = data_df['id'].apply(
+        #         lambda x: (x == node) or (node in ancestors_of_nodes[x])
+        #     )
+
+        return data_df
+
+    @staticmethod
+    def get_transitively_closed_graph() -> nx.DiGraph:
+        """
+        Create a directed graph, compute its transitive closure, and return it.
+
+        Returns:
+            g (nx.DiGraph): A transitively closed directed graph.
+        """
+        g = nx.DiGraph()
+
+        for node in ChebiMockOntology.get_nodes():
+            g.add_node(node, **{"smiles": "test_smiles_placeholder"})
+
+        g.add_edges_from(ChebiMockOntology.get_edges_of_transitive_closure_graph())
+
+        return g
+
+
 class GOUniProtMockData(MockOntologyGraphData):
     """
     A mock ontology representing a simplified version of the Gene Ontology (GO) structure with nodes and edges

Original file line number	Diff line number	Diff line change
`@@ -1 +1 @@`
`1`		`-from .base import XYBaseDataModule, _DynamicDataset`
	`1`	`+from .base import XYBaseDataModule, _DynamicDataset`