Merge branch '254_induced_eic_associate_order' into 'dev'

Associate induced EICs before clearing gap-fill features (#254)

See merge request mass-spectrometry/corems!238

Author: kheal

corems/mass_spectra/calc/lc_calc.py

@@ -5769,22 +5769,22 @@ def process_consensus_features(self, load_representatives=True, perform_gap_fill
                         eics_mz.append(None)
                 self.induced_mass_features_dataframe['_eic_mz'] = eics_mz
 
-            # Clear mass features from samples to free memory
-            for sample_name in self.samples:
-                self._lcms[sample_name].induced_mass_features = {}
-        
-        # Associate EICs with mass features if they were loaded
-        # This must happen after all operations complete to work on the actual sample objects
+        # Associate EICs while induced feature objects still exist (before any clear).
+        # Must run after the pipeline so sample.eics is populated on the main process.
         if gather_eics:
             print("\nAssociating EICs with mass features:")
             from tqdm import tqdm
-            
+
             for sample_id in tqdm(range(len(self.samples)), unit="sample", ncols=80):
                 sample = self[sample_id]
                 if sample.eics:  # Only if EICs were loaded
-                    # Associate EICs with regular mass features
                     sample.associate_eics_with_mass_features(induced=False)
-                    # Associate EICs with induced mass features
                     sample.associate_eics_with_mass_features(induced=True)
-                
+
+        # Drop induced feature objects to free memory. EICs remain on sample.eics
+        # (and _eic_mz on the induced dataframe) for plotting/lookup.
+        if perform_gap_filling:
+            for sample_name in self.samples:
+                self._lcms[sample_name].induced_mass_features = {}
+
         return results

tests/test_lcms_collection.py

@@ -307,12 +307,63 @@ def test_lcms_collection_gap_filling(lcms_collection):
 
     # Sample 3 should have induced features (started with 0, all 50 clusters are missing)
     assert sample_3_induced == 50, "Sample 3 should have 50 induced mass features (one for each cluster)"
-    
-    # By design, individual sample objects should have empty induced_mass_features dict
-    # because they are collected into the induced_mass_features_dataframe
-    assert len(lcms_collection[0].induced_mass_features) == 0
-    assert len(lcms_collection[1].induced_mass_features) == 0
-    assert len(lcms_collection[2].induced_mass_features) == 0
+
+    # Induced feature *objects* are always cleared after gap-fill (memory).
+    for i in range(len(lcms_collection)):
+        assert len(lcms_collection[i].induced_mass_features) == 0, (
+            f"Sample {i} should clear induced_mass_features after gap-fill"
+        )
+
+    # EICs are separate: gap-fill + gather_eics leaves chromatograms on sample.eics
+    # keyed by m/z (including induced feature m/z).
+    sample3 = lcms_collection[2]
+    assert sample3.eics is not None and len(sample3.eics) > 0, (
+        "Sample 3 should retain EICs on sample.eics after gap-fill with gather_eics=True"
+    )
+    if "_eic_mz" in induced_df.columns:
+        induced_mzs = induced_df.loc[induced_df["sample_id"] == 2, "_eic_mz"].dropna()
+        assert len(induced_mzs) > 0
+        # At least some induced m/z values should resolve in sample.eics
+        from corems.mass_spectra.factory.lc_class import LCMSBase
+        matched = 0
+        for mz in induced_mzs:
+            if mz in sample3.eics:
+                matched += 1
+            elif hasattr(sample3, "get_eic_mz_for_mass_feature"):
+                key = sample3.get_eic_mz_for_mass_feature(float(mz))
+                if key is not None:
+                    matched += 1
+        assert matched > 0, (
+            "Induced feature m/z values should be present in sample.eics after gap-fill"
+        )
+
+
+def test_lcms_collection_gap_fill_clears_induced_without_gather_eics(lcms_collection):
+    """Without gather_eics, induced objects are still cleared after gap-fill."""
+    lcms_collection = copy.deepcopy(lcms_collection)
+    if not lcms_collection.rt_alignment_attempted:
+        lcms_collection.align_lcms_objects()
+    lcms_collection.add_consensus_mass_features()
+
+    lcms_collection.process_consensus_features(
+        load_representatives=False,
+        perform_gap_filling=True,
+        add_ms1=False,
+        add_ms2=False,
+        molecular_formula_search=False,
+        ms2_spectral_search=False,
+        spectral_lib=False,
+        molecular_metadata=None,
+        gather_eics=False,
+        keep_raw_data=False,
+    )
+
+    assert lcms_collection.induced_mass_features_dataframe is not None
+    assert len(lcms_collection.induced_mass_features_dataframe) > 0
+    for i in range(len(lcms_collection)):
+        assert len(lcms_collection[i].induced_mass_features) == 0, (
+            f"Sample {i} should clear induced_mass_features after gap-fill"
+        )
 
 
 def test_lcms_collection_pivot_table(lcms_collection):