Add 3Di sequences for existing test structure

tests/structure/data/alphabet/README.rst

@@ -0,0 +1,16 @@
+Structural alphabet sequences
+==============================
+
+This directory contains structural alphabet sequences for the test structure files
+from the `tests/structure/data/` directory, generated with the respective reference
+implementation.
+
+3Di sequences
+-------------
+
+The 3Di sequences in `i3d.fasta` were generated with `foldseek`:
+
+.. code-block:: console
+
+    $ foldseek createdb --chain-name-mode 1 tests/structure/data/*.cif /tmp/biotite_3di
+    $ foldseek convert2fasta /tmp/biotite_3di tests/structure/data/alphabet/i3d.fasta

tests/structure/data/ids.txt

@@ -16,6 +16,4 @@
 5eil
 4p5j
 1crr
-7gsa
-8crb
-3bww
+7gsa

tests/structure/test_i3d.py

@@ -0,0 +1,69 @@
+import re
+from pathlib import Path
+import pytest
+import biotite.structure as struc
+import biotite.structure.io.pdbx as pdbx
+import biotite.sequence.io.fasta as fasta
+import biotite.structure.alphabet as strucalph
+import biotite.structure.io.pdb as pdb
+from tests.util import data_dir
+
+
+def _get_ref_3di_sequence(pdb_id, chain_id):
+    """
+    Get the reference 3di sequence for the first model of the structure with the given
+    PDB ID and chain ID.
+    """
+    ref_3di_file = fasta.FastaFile.read(
+        Path(data_dir("structure")) / "alphabet" / "i3d.fasta"
+    )
+    for header, seq_string in ref_3di_file.items():
+        # The first model of a structure is also the first sequence to appear
+        # and thus to be matched
+        if re.match(rf"^{pdb_id}(_MODEL_\d+)?_{chain_id}", header):
+            ref_3di_sequence = strucalph.I3DSequence(seq_string)
+            break
+    else:
+        raise ValueError(
+            f"Reference 3Di sequence not found for {pdb_id} chain {chain_id}"
+        )
+    return ref_3di_sequence
+
+
+@pytest.mark.parametrize(
+    "path", Path(data_dir("structure")).glob("*.bcif"), ids=lambda path: path.stem
+)
+def test_to_3di(path):
+    """
+    Check if the 3di sequence of a chain is correctly generated, by comparing the result
+    to a reference sequence generated with *foldseek*.
+    """
+    pdbx_file = pdbx.BinaryCIFFile.read(path)
+    atoms = pdbx.get_structure(pdbx_file, model=1)
+    atoms = atoms[struc.filter_amino_acids(atoms)]
+    if len(atoms) == 0:
+        # Skip empty structures
+        return
+    test_3di, chain_starts = strucalph.to_3di(atoms)
+
+    ref_3di = [
+        _get_ref_3di_sequence(path.stem, chain_id)
+        for chain_id in atoms.chain_id[chain_starts]
+    ]
+
+    for (test, ref, chain_id) in zip(test_3di, ref_3di, atoms.chain_id[chain_starts]):
+        assert str(test) == str(ref), f"3Di sequence of chain {chain_id} does not match"
+
+
+def test_missing():
+    """
+    Test if missing or non-peptide residues within a chain are correctly handled.
+    """
+    pass
+
+
+def test_empty():
+    """
+    Test if an empty structure is correctly handled.
+    """
+    pass