[ENH] Replace PFOA with 5NU7 and 1BRQ

examples/aptanet_tutorial.ipynb

@@ -19,7 +19,7 @@
     "\n",
     "- **pairs_to_features**: helper that converts `(aptamer_sequence, protein_sequence)` pairs into feature vectors using k-mer + PSeAAC.\n",
     "- **SkorchAptaNet**: a PyTorch MLP wrapped in Skorch for binary classification.\n",
-    "- **load_pfoa_structure**: helper to load the PFOA molecule structure from PDB file."
+    "- **load_1gnh_structure**: helper to load the GNH molecule structure from PDB file."
    ]
   },
   {

pyaptamer/data/tests/test_loader.py

@@ -6,7 +6,7 @@
 
 # dataset paths relative to pyaptamer root
 DATA_PATHS = [
-    "datasets/data/pfoa.pdb",
+    "datasets/data/5nu7.pdb",
     "datasets/data/1gnh.pdb",
 ]
 

pyaptamer/datasets/__init__.py

@@ -1,24 +1,31 @@
 """Contains datasets along with their loaders."""
 
-from pyaptamer.datasets._loaders._aptacom_loader import (
+from pyaptamer.datasets._loaders import (
+    load_1brq,
+    load_1brq_structure,
+    load_1gnh,
+    load_1gnh_structure,
+    load_5nu7,
+    load_5nu7_structure,
     load_aptacom_full,
     load_aptacom_x_y,
+    load_csv_dataset,
+    load_from_rcsb,
+    load_hf_dataset,
 )
-from pyaptamer.datasets._loaders._csv_loader import load_csv_dataset
-from pyaptamer.datasets._loaders._hf_loader import load_hf_dataset
-from pyaptamer.datasets._loaders._one_gnh import load_1gnh, load_1gnh_structure
-from pyaptamer.datasets._loaders._online_databank import load_from_rcsb
-from pyaptamer.datasets._loaders._pfoa import load_pfoa, load_pfoa_structure
 
 __all__ = [
     "load_aptacom_full",
     "load_aptacom_x_y",
     "load_csv_dataset",
     "load_hf_dataset",
-    "load_pfoa",
-    "load_pfoa_structure",
+    "load_from_rcsb",
+    "mol_loader",
+    "structure_loader",
+    "load_1brq",
+    "load_1brq_structure",
+    "load_5nu7",
+    "load_5nu7_structure",
     "load_1gnh",
     "load_1gnh_structure",
-    "load_from_rcsb",
-    "load_csv_dataset",
 ]

pyaptamer/datasets/_loaders/_pfoa.py → pyaptamer/datasets/_loaders/_1brq.py

@@ -1,43 +1,43 @@
-__author__ = ["satvshr", "fkiraly"]
-__all__ = ["load_pfoa", "load_pfoa_structure"]
+__author__ = "satvshr"
+__all__ = ["load_1brq", "load_1brq_structure"]
 
 import os
 
 
-def load_pfoa():
-    """Load the PFOA molecule as a MoleculeLoader.
+def load_1brq():
+    """Load the 1brq molecule as a MoleculeLoader.
 
     Returns
     -------
     loader : MoleculeLoader
-        A MoleculeLoader object representing the PFOA molecule.
+        A MoleculeLoader object representing the 1brq molecule.
     """
     from pyaptamer.data.loader import MoleculeLoader
 
-    pdb_path = os.path.join(os.path.dirname(__file__), "..", "data", "pfoa.pdb")
+    pdb_path = os.path.join(os.path.dirname(__file__), "..", "data", "1brq.pdb")
 
     return MoleculeLoader(pdb_path)
 
 
-def load_pfoa_structure(pdb_path=None):
+def load_1brq_structure(pdb_path=None):
     """
-    Load the PFOA molecule from a PDB file using Biopython.
+    Load the 1brq molecule from a PDB file using Biopython.
 
     Parameters
     ----------
     pdb_path : str, optional
         Path to the PDB file. If not provided, the function uses the default path
-        located in the '../data/pfoa.pdb' relative to the current file.
+        located in the '../data/1brq.pdb' relative to the current file.
 
     Returns
     -------
     structure : Bio.PDB.Structure.Structure
-        A Biopython Structure object representing the PFOA molecule.
+        A Biopython Structure object representing the 1BRQ molecule.
     """
     from Bio.PDB import PDBParser
 
-    pdb_path = os.path.join(os.path.dirname(__file__), "..", "data", "pfoa.pdb")
+    pdb_path = os.path.join(os.path.dirname(__file__), "..", "data", "1brq.pdb")
 
     parser = PDBParser(QUIET=True)
-    structure = parser.get_structure("PFOA", pdb_path)
+    structure = parser.get_structure("1brq", pdb_path)
     return structure

pyaptamer/datasets/_loaders/_one_gnh.py → pyaptamer/datasets/_loaders/_1gnh.py

@@ -32,7 +32,7 @@ def load_1gnh_structure(pdb_path=None):
     Returns
     -------
     structure : Bio.PDB.Structure.Structure
-        A Biopython Structure object representing the PFOA molecule.
+        A Biopython Structure object representing the 1GNH molecule.
     """
     from Bio.PDB import PDBParser
 

pyaptamer/datasets/_loaders/_5nu7.py

@@ -0,0 +1,43 @@
+__author__ = "satvshr"
+__all__ = ["load_5nu7", "load_5nu7_structure"]
+
+import os
+
+
+def load_5nu7():
+    """Load the 5nu7 molecule as a MoleculeLoader.
+
+    Returns
+    -------
+    loader : MoleculeLoader
+        A MoleculeLoader object representing the 5nu7 molecule.
+    """
+    from pyaptamer.data.loader import MoleculeLoader
+
+    pdb_path = os.path.join(os.path.dirname(__file__), "..", "data", "5nu7.pdb")
+
+    return MoleculeLoader(pdb_path)
+
+
+def load_5nu7_structure(pdb_path=None):
+    """
+    Load the 5nu7 molecule from a PDB file using Biopython.
+
+    Parameters
+    ----------
+    pdb_path : str, optional
+        Path to the PDB file. If not provided, the function uses the default path
+        located in the '../data/5nu7.pdb' relative to the current file.
+
+    Returns
+    -------
+    structure : Bio.PDB.Structure.Structure
+        A Biopython Structure object representing the 5NU7 molecule.
+    """
+    from Bio.PDB import PDBParser
+
+    pdb_path = os.path.join(os.path.dirname(__file__), "..", "data", "5nu7.pdb")
+
+    parser = PDBParser(QUIET=True)
+    structure = parser.get_structure("5nu7", pdb_path)
+    return structure

pyaptamer/datasets/_loaders/__init__.py

@@ -1,22 +1,35 @@
 """Loaders for different data structures."""
 
+from pyaptamer.datasets._loaders._1brq import (
+    load_1brq,
+    load_1brq_structure,
+)
+from pyaptamer.datasets._loaders._1gnh import (
+    load_1gnh,
+    load_1gnh_structure,
+)
+from pyaptamer.datasets._loaders._5nu7 import (
+    load_5nu7,
+    load_5nu7_structure,
+)
 from pyaptamer.datasets._loaders._aptacom_loader import (
     load_aptacom_full,
     load_aptacom_x_y,
 )
 from pyaptamer.datasets._loaders._csv_loader import load_csv_dataset
 from pyaptamer.datasets._loaders._hf_loader import load_hf_dataset
-from pyaptamer.datasets._loaders._one_gnh import load_1gnh_structure
-from pyaptamer.datasets._loaders._pfoa import load_pfoa_structure
+from pyaptamer.datasets._loaders._online_databank import load_from_rcsb
 
 __all__ = [
-    "load_pfoa_structure",
-    "load_1gnh_structure",
     "load_aptacom_full",
     "load_aptacom_x_y",
     "load_csv_dataset",
+    "load_from_rcsb",
     "load_hf_dataset",
-    "load_pfoa_structure",
     "load_1gnh",
     "load_1gnh_structure",
+    "load_1brq",
+    "load_1brq_structure",
+    "load_5nu7",
+    "load_5nu7_structure",
 ]