From 21b1f101e4947ffb880427540b34074399809ecf Mon Sep 17 00:00:00 2001
From: JochenSiegWork <135010976+JochenSiegWork@users.noreply.github.com>
Date: Fri, 27 Sep 2024 11:59:26 +0200
Subject: [PATCH] mol2morgan_fingerprint: fix bug of mismatched indices after
 folding (#93)

---
 molpipeline/mol2any/mol2morgan_fingerprint.py        |  5 ++---
 .../test_mol2any/test_mol2morgan_fingerprint.py      | 12 ++----------
 2 files changed, 4 insertions(+), 13 deletions(-)

diff --git a/molpipeline/mol2any/mol2morgan_fingerprint.py b/molpipeline/mol2any/mol2morgan_fingerprint.py
index 79fa46c1..36fca193 100644
--- a/molpipeline/mol2any/mol2morgan_fingerprint.py
+++ b/molpipeline/mol2any/mol2morgan_fingerprint.py
@@ -154,8 +154,7 @@ def _explain_rdmol(self, mol_obj: RDKitMol) -> dict[int, list[tuple[int, int]]]:
         fp_generator = self._get_fp_generator()
         additional_output = AllChem.AdditionalOutput()
         additional_output.AllocateBitInfoMap()
-        _ = fp_generator.GetSparseFingerprint(
-            mol_obj, additionalOutput=additional_output
-        )
+        # using the dense fingerprint here, to get indices after folding
+        _ = fp_generator.GetFingerprint(mol_obj, additionalOutput=additional_output)
         bit_info = additional_output.GetBitInfoMap()
         return bit_info
diff --git a/tests/test_elements/test_mol2any/test_mol2morgan_fingerprint.py b/tests/test_elements/test_mol2any/test_mol2morgan_fingerprint.py
index 6fae46b4..2b765949 100644
--- a/tests/test_elements/test_mol2any/test_mol2morgan_fingerprint.py
+++ b/tests/test_elements/test_mol2any/test_mol2morgan_fingerprint.py
@@ -130,16 +130,8 @@ def test_setter_getter_error_handling(self) -> None:
         self.assertRaises(ValueError, mol_fp.set_params, **params)
 
     def test_bit2atom_mapping(self) -> None:
-        """Test that the mapping from bits to atom weights works as intended.
-
-        Notes
-        -----
-        lower n_bit values, e.g. 2048, will lead to a bit clash during folding,
-        for the test smiles "NCCOCCCC(=O)O".
-        We want no folding clashes in this test to check the correct length
-        of the bit-to-atom mapping.
-        """
-        n_bits = 2100
+        """Test that the mapping from bits to atom weights works as intended."""
+        n_bits = 2048
         sparse_morgan = MolToMorganFP(radius=2, n_bits=n_bits, return_as="sparse")
         dense_morgan = MolToMorganFP(radius=2, n_bits=n_bits, return_as="dense")
         explicit_bit_vect_morgan = MolToMorganFP(