Always write an optics block

Author: j-c-c

src/aspire/source/image.py

@@ -1311,31 +1311,26 @@ def _prepare_relion_optics_blocks(metadata):
             "_rlnImageDimensionality",
         ]
 
-        # We only write a data_optics block when every required field is present.
-        required_fields = [
-            "_rlnSphericalAberration",
-            "_rlnVoltage",
-            "_rlnAmplitudeContrast",
-            "_rlnImageSize",
-            "_rlnImageDimensionality",
-        ]
-        pixel_fields = ["_rlnImagePixelSize", "_rlnMicrographPixelSize"]
+        # Some optics group fields might not always be present, but are necessary
+        # for reading the file in Relion. We ensure these fields exist and populate
+        # with a dummy value if not.
+        n_rows = len(metadata["_rlnImageName"])
 
-        has_required = all(field in metadata for field in required_fields)
-        has_pixel_field = any(field in metadata for field in pixel_fields)
+        def _ensure_column(field, value):
+            if field not in metadata:
+                logger.warning(
+                    f"Optics field {field} not found, populating with default value {value}"
+                )
+                metadata[field] = np.full(n_rows, value)
 
-        if not (has_required and has_pixel_field):
-            # Optics metadata incomplete, fall back to legacy single block.
-            logger.warning(
-                "Optics metadata incomplete, writing only data_particles block."
-            )
-            return None, metadata
+        _ensure_column("_rlnSphericalAberration", 2.0)
+        _ensure_column("_rlnVoltage", 300)
+        _ensure_column("_rlnAmplitudeContrast", 0.1)
 
         # Restrict to the optics columns that are actually present on this source.
         optics_value_fields = [
             field for field in all_optics_fields if field in metadata
         ]
-        n_rows = len(metadata["_rlnImageName"])
 
         # Map each unique optics tuple to a 1-based group ID in order encountered.
         group_lookup = OrderedDict()

src/aspire/source/relion.py

@@ -168,7 +168,8 @@ def __init__(
                 f"Found partially populated CTF Params."
                 f"  To automatically populate CTFFilters provide {CTF_params}"
             )
-
+            self.unique_filters = [IdentityFilter()]
+            self.filter_indices = np.zeros(self.n, dtype=int)
         # If no CTF info in STAR, we initialize the filter values of metadata with default values
         else:
             self.unique_filters = [IdentityFilter()]

tests/test_coordinate_source.py

@@ -526,7 +526,7 @@ def testSave(self):
         # load saved particle stack
         saved_star = StarFile(star_path)
         # we want to read the saved mrcs file from the STAR file
-        image_name_column = saved_star.get_block_by_index(0)["_rlnImageName"]
+        image_name_column = saved_star.get_block_by_index(1)["_rlnImageName"]
         # we're reading a string of the form 0000X@mrcs_path.mrcs
         _particle, mrcs_path = image_name_column[0].split("@")
         saved_mrcs_stack = mrcfile.open(os.path.join(self.data_folder, mrcs_path)).data
@@ -537,15 +537,28 @@ def testSave(self):
         self.assertEqual(
             list(saved_star["particles"].keys()),
             [
-                "_rlnImagePixelSize",
+                "_rlnOpticsGroup",
                 "_rlnSymmetryGroup",
                 "_rlnImageName",
                 "_rlnCoordinateX",
                 "_rlnCoordinateY",
+            ],
+        )
+
+        self.assertEqual(
+            list(saved_star["optics"].keys()),
+            [
+                "_rlnOpticsGroup",
+                "_rlnOpticsGroupName",
+                "_rlnImagePixelSize",
+                "_rlnSphericalAberration",
+                "_rlnVoltage",
+                "_rlnAmplitudeContrast",
                 "_rlnImageSize",
                 "_rlnImageDimensionality",
             ],
         )
+
         # assert that all the correct coordinates were saved
         for i in range(10):
             self.assertEqual(

tests/test_relion_source.py

@@ -70,6 +70,7 @@ def test_prepare_relion_optics_blocks_warns(caplog):
         "_rlnImagePixelSize": np.array([1.234]),
         "_rlnImageSize": np.array([32]),
         "_rlnImageDimensionality": np.array([2]),
+        "_rlnImageName": np.array(["[email protected]"]),
     }
 
     caplog.clear()
@@ -78,9 +79,21 @@ def test_prepare_relion_optics_blocks_warns(caplog):
             metadata.copy()
         )
 
-    assert optics_block is None
-    assert particle_block == metadata
-    assert "Optics metadata incomplete" in caplog.text
+    # We should get and optics block
+    assert optics_block is not None
+
+    # Verify defaults were injected.
+    np.testing.assert_allclose(optics_block["_rlnImagePixelSize"], [1.234])
+    np.testing.assert_array_equal(optics_block["_rlnImageSize"], [32])
+    np.testing.assert_array_equal(optics_block["_rlnImageDimensionality"], [2])
+    np.testing.assert_allclose(optics_block["_rlnVoltage"], [300.0])
+    np.testing.assert_allclose(optics_block["_rlnSphericalAberration"], [2.0])
+    np.testing.assert_allclose(optics_block["_rlnAmplitudeContrast"], [0.1])
+
+    # Caplog should contain the warnings about the three missing fields.
+    assert "Optics field _rlnSphericalAberration not found" in caplog.text
+    assert "Optics field _rlnVoltage not found" in caplog.text
+    assert "Optics field _rlnAmplitudeContrast not found" in caplog.text
 
 
 def test_pixel_size(caplog):