Support falcon datasets with latest micromanager (#237)

* use "DefaultXYStage" to get stage positions

* read new MM tile creator strings

* style

* fix tests

* simplify getting stage position name

* use regex for HCS label parsing

* bugfix

* more strict label matching

* add DefaultXYStage to ndtiff stage_positions metadata

* modify getting stage positions

* add tests that stage position is present in metadata

---------

Co-authored-by: Ivan Ivanov <[email protected]>

iohub/convert.py

@@ -200,10 +200,10 @@ def _get_position_coords(self):
         if not self.reader.stage_positions:
             raise ValueError("Stage positions not available.")
         for idx, pos in enumerate(self.reader.stage_positions):
-            stage_pos = (
-                pos.get("XYStage") or pos.get("XY") or pos.get("XY Stage")
-            )
-            if stage_pos is None:
+            try:
+                xy_stage = pos["DefaultXYStage"]
+                stage_pos = pos[xy_stage]
+            except KeyError:
                 raise ValueError(
                     f"Stage position is not available for position {idx}"
                 )

iohub/mm_fov.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import re
 from pathlib import Path
 
 from xarray import DataArray
@@ -109,42 +110,53 @@ def stage_positions(self, value):
     @property
     def hcs_position_labels(self):
         """Parse plate position labels generated by the HCS position generator,
-        e.g. 'A1-Site_0' or '1-Pos000_000', and split into row, column, and
-        FOV names.
+        e.g. 'A1-Site_0', '1-Pos000_000', or 'Pos-9-000_000', and split into
+        row, column, and FOV names.
 
         Returns
         -------
         list[tuple[str, str, str]]
-            FOV name paths, e.g. ('A', '1', '0') or ('0', '1', '000000')
+            FOV name paths, e.g. ('A', '1', '0'), ('0', '1', '000000'), or
+            ('0', '9', '000000').
         """
         if not self.stage_positions:
             raise ValueError("Stage position metadata not available.")
+
         try:
-            # Look for "'A1-Site_0', 'H12-Site_1', ... " format
-            labels = [
-                pos["Label"].split("-Site_") for pos in self.stage_positions
-            ]
-            return [(well[0], well[1:], fov) for well, fov in labels]
-        except Exception:
-            try:
-                # Look for "'1-Pos000_000', '2-Pos000_001', ... "
-                # and split into ('1', '000_000'), ...
-                labels = [
-                    pos["Label"].split("-Pos") for pos in self.stage_positions
-                ]
-                # remove underscore from FOV name, i.e. '000_000'
-                # collect all wells in row '0' so output is
-                # ('0', '1', '000000')
-                return [
-                    ("0", col, fov.replace("_", "")) for col, fov in labels
-                ]
-            except Exception:
-                labels = [pos.get("Label") for pos in self.stage_positions]
-                raise ValueError(
-                    "HCS position labels are in the format of "
-                    "'A1-Site_0', 'H12-Site_1', or '1-Pos000_000' "
-                    f"Got labels {labels}"
-                )
+            labels = [pos["Label"] for pos in self.stage_positions]
+        except KeyError:
+            raise ValueError("Stage positions do not have labels.")
+
+        # See https://chatgpt.com/share/e/670097cc-2854-8008-bd33-b54cad7c99b9
+        pattern = re.compile(
+            r"([A-Z])(\d+)-Site_(\d+)|"
+            r"Pos-(\d+)-(\d+)_(\d+)|"
+            r"(\d+)-Pos(\d+)_(\d+)"
+        )
+        row_col_fov = []
+        for label in labels:
+            if (match := re.match(pattern, label)) is not None:
+                if match.group(1):  # "A1-Site_0" case
+                    row_col_fov.append(
+                        (match.group(1), match.group(2), match.group(3))
+                    )
+                elif match.group(4):  # "Pos-5-000_005" case
+                    row_col_fov.append(
+                        ("0", match.group(4), match.group(5) + match.group(6))
+                    )
+                elif match.group(7):  # "1-Pos000_000" case
+                    row_col_fov.append(
+                        ("0", match.group(7), match.group(8) + match.group(9))
+                    )
+
+        if not row_col_fov:
+            raise ValueError(
+                "HCS position labels are in the format of "
+                "'A1-Site_0', 'H12-Site_1', or '1-Pos000_000', "
+                f"or 'Pos-1-000_000'. Got labels {labels}"
+            )
+
+        return row_col_fov
 
     @property
     def zyx_scale(self) -> tuple[float, float, float]:

iohub/ndtiff.py

@@ -167,7 +167,9 @@ def _get_summary_metadata(self):
                         "YPosition_um_Intended",
                     ]
                 ):
-                    position_metadata[img_metadata["Core-XYStage"]] = (
+                    xy_stage = img_metadata["Core-XYStage"]
+                    position_metadata["DefaultXYStage"] = xy_stage
+                    position_metadata[xy_stage] = (
                         img_metadata["XPosition_um_Intended"],
                         img_metadata["YPosition_um_Intended"],
                     )

tests/mmstack/test_mmstack.py

@@ -68,6 +68,8 @@ def test_mmstack_metadata(ome_tiff):
         assert isinstance(mmstack.micromanager_metadata, dict)
         assert mmstack.micromanager_metadata["Summary"]
         assert mmstack.micromanager_summary
+        if mmstack.stage_positions:
+            assert "DefaultXYStage" in mmstack.stage_positions[0]
 
 
 def test_fov_axes_names(ome_tiff):

tests/test_ndtiff.py

@@ -32,6 +32,8 @@ def test_dataset_metadata(ndtiff_dataset):
         assert isinstance(dataset.micromanager_metadata, dict)
         assert dataset.micromanager_metadata["Summary"]
         assert isinstance(dataset.micromanager_summary, dict)
+        if dataset.stage_positions:
+            assert "DefaultXYStage" in dataset.stage_positions[0]
 
 
 @pytest.mark.parametrize("ndtiff_v2", ndtiff_v2_datasets)