proposal for simplified API for tiled acquisition

examples/hcs_zarr.py

@@ -37,8 +37,11 @@
     # create and write to positions
     for i, (row, col, fov) in enumerate(position_list):
         position = dataset.create_position(row, col, fov)
-        position["0"] = i*np.ones((2, 3, 5, 32, 32)).astype(np.uint16) 
+        position["0"] = i * np.ones((2, 3, 5, 32, 32)).astype(np.uint16)
         # 2 timepoints, 3 channels, 5 z-slices, 32x32 image
+
+        #position.img or position.data is a more intuitive way to write and read data. 
+        # Can we implement this?
     # print dataset summary
     dataset.print_tree()
 # Try viewing the hcs.zarr dataset in napari.

examples/singleFOV_zarr.py

@@ -23,6 +23,8 @@
     store_path, layout="fov", mode="a", channel_names=["DAPI", "GFP"]
 ) as dataset:
     dataset["img"] = tczyx
+    #dataset.img or dataset.data is a more intuitive way to write and read data. 
+    # Can we implement this?
 
 # %%
 # Opening in read-only mode prevents writing

examples/tiles_zarr.py

@@ -0,0 +1,41 @@
+# %%
+# This script writes a tiled (multi-FOV) zarr dataset.
+# It is not spcified by the NGFF specification, but is a common use case.
+# We repurpose the HCS layout from the NGFF specification to make multi-position data to view with napari-ome-zarr.
+
+
+import os
+
+import numpy as np
+
+from iohub.ngff import open_ome_zarr
+
+# %%
+# Set storage path
+
+store_path = f'{os.path.expanduser("~/")}tiles.zarr'
+
+# %%
+# Write 5D data to multiple positions.
+# While the NGFF specification allows for arbitrary names,
+# the ome-zarr-py library (thus the napari-ome-zarr plugin)
+# only load arrays with name '0'.
+
+position_list = ((row, col) for row in ("row_0", "row_1") for col in range(10))
+
+with open_ome_zarr(
+    store_path,
+    layout="tiles",
+    mode="w",
+    channel_names=["DAPI", "GFP", "Brightfield"],
+) as dataset:
+    # create and write to positions
+    for i, (row, col) in enumerate(position_list):
+        position = dataset.create_position(row, col)
+        # Following usage (set the data property) is more intuitive. Can we implement this?
+        # position.data = i*np.ones((2, 3, 5, 32, 32)).astype(np.uint16) 
+        position["0"] = i*np.ones((2, 3, 5, 32, 32)).astype(np.uint16)
+        # 2 timepoints, 3 channels, 5 z-slices, 32x32 image at each position
+    # print dataset summary
+    dataset.print_tree()
+# Try viewing the tiles.zarr dataset in napari or each position individually.