fix mol props for as per node data cls

Author: aditya0by0

chebai_graph/preprocessing/datasets/chebi.py

@@ -376,14 +376,14 @@ def load_processed_data_from_file(self, filename: str) -> list[dict]:
             n_atom_node_properties, n_fg_node_properties, n_graph_node_properties
         )
         rank_zero_info(
-            f"Finished loading dataset from properties.\nEncoding lengths: {prop_lengths}\n"
-            f"Properties Categories:\n{pformat(props_categories)}"
+            f"\nFinished loading dataset from properties.\nEncoding lengths: {prop_lengths}\n\n"
+            f"Properties Categories:\n{pformat(props_categories)}\n\n"
             f"n_atom_node_properties: {n_atom_node_properties}, "
             f"n_fg_node_properties: {n_fg_node_properties}, "
             f"n_bond_properties: {n_bond_properties}, "
-            f"n_graph_node_properties: {n_graph_node_properties}\n"
+            f"n_graph_node_properties: {n_graph_node_properties}\n\n"
             f"Use following values for given parameters for model configuration: \n\t"
-            f"in_channels: {n_node_properties}, edge_dim: {n_bond_properties}, n_molecule_properties: 0"
+            f"in_channels: {n_node_properties}, edge_dim: {n_bond_properties}, n_molecule_properties: 0\n"
         )
 
         for property in self.properties:
@@ -449,7 +449,7 @@ def _merge_props_into_base(
         atom_offset, fg_offset, graph_offset = 0, 0, 0
 
         for property in self.properties:
-            property_values = row[f"{property.name}"]
+            property_values = row[f"{property.name}"].to(dtype=torch.float32)
             if isinstance(property_values, torch.Tensor):
                 if len(property_values.size()) == 0:
                     property_values = property_values.unsqueeze(0)
@@ -482,7 +482,7 @@ def _merge_props_into_base(
 
             elif isinstance(property, MoleculeProperty):
                 x[is_graph_node, graph_offset : graph_offset + enc_len] = (
-                    property_values[is_graph_node]
+                    property_values
                 )
                 graph_offset += enc_len
 
@@ -505,6 +505,7 @@ def _merge_props_into_base(
             x=x,
             edge_index=geom_data.edge_index,
             edge_attr=edge_attr,
+            molecule_attr=torch.empty((1, 0)),  # empty as not used for this class
         )
 
 

chebai_graph/preprocessing/properties/properties.py

@@ -275,6 +275,12 @@ class RDKit2DNormalized(MoleculeProperty):
     def __init__(self, encoder: PropertyEncoder | None = None) -> None:
         super().__init__(encoder or AsIsEncoder(self))
         self.generator_normalized = rdNormalizedDescriptors.RDKit2DNormalized()
+        # Create a dummy molecule (e.g., methane) to extract the length of descriptor vector
+        dummy_mol = Chem.MolFromSmiles("C")
+        descr_values = self.generator_normalized.processMol(
+            dummy_mol, Chem.MolToSmiles(dummy_mol)
+        )
+        self.encoder.set_encoding_length(len(descr_values) - 1)
 
     def get_property_value(self, mol: Chem.rdchem.Mol) -> list[np.ndarray]:
         """

chebai_graph/preprocessing/property_encoder.py

@@ -243,8 +243,16 @@ def encode(self, token: float | int | None) -> torch.Tensor:
             Tensor of shape (1,) containing the input value or zero.
         """
         if token is None:
-            return torch.tensor([0])
-        return torch.tensor([token])
+            return torch.zeros(1, self.get_encoding_length())
+        assert (
+            len(token) == self.get_encoding_length()
+        ), "Length of token should be equal to encoding length"
+        # return torch.tensor([token]) # token is an ndarray, no need to create list of ndarray due to below warning
+        # UserWarning: Creating a tensor from a list of numpy.ndarrays is extremely slow.
+        # Please consider converting the list to a single numpy.ndarray with numpy.array() before converting to a tensor.
+        # (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\pytorch\torch\csrc\utils\tensor_new.cpp:257.)
+        # ----- fix: for above warning
+        return torch.tensor(token).unsqueeze(0)  # shape: (1, len(token))
 
 
 class BoolEncoder(PropertyEncoder):