use proper datasets

YeoLab · Apr 26, 2024 · 5038087 · 5038087
1 parent e981cdf
commit 5038087
Show file tree

Hide file tree

Showing 45 changed files with 364 additions and 588 deletions.
diff --git a/bento/_utils.py b/bento/_utils.py
@@ -38,9 +38,7 @@ def filter_by_gene(
         .table is updated to remove genes with low expression.
     """
     gene_filter = (sdata.table.X >= min_count).sum(axis=0) > 0
-    table_transform = sdata.table.attrs
     filtered_table = sdata.table[:, gene_filter]
-    filtered_table.attrs = table_transform
 
     filtered_genes = list(sdata.table.var_names.difference(filtered_table.var_names))
     points = get_points(sdata, points_key=points_key, astype="pandas", sync=False)

diff --git a/bento/datasets/_datasets.py b/bento/datasets/_datasets.py
@@ -106,5 +106,11 @@ def sample_data():
     ref = files(__package__) / "merfish_sample.zarr"
     with as_file(ref) as path:
         sdata = sd.read_zarr(path)
-        sdata = prep(sdata, points_key="transcripts", instance_key="feature_name", shape_keys=["cell_boundaries", "nucleus_boundaries"])
-        return sdata
+        sdata = prep(
+            sdata,
+            points_key="transcripts",
+            instance_key="cell_boundaries",
+            feature_key="feature_name",
+            shape_keys=["cell_boundaries", "nucleus_boundaries"],
+        )
+        return sdata
diff --git a/bento/datasets/merfish_sample.zarr/points/transcripts/points.parquet/part.0.parquet b/bento/datasets/merfish_sample.zarr/points/transcripts/points.parquet/part.0.parquet
diff --git a/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/Index/.zarray b/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/Index/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        15
+        14
     ],
     "compressor": {
         "blocksize": 0,
@@ -19,7 +19,7 @@
     ],
     "order": "C",
     "shape": [
-        15
+        14
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/Index/0 b/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/Index/0
diff --git a/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/coords/.zarray b/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/coords/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        1443,
+        1224,
         2
     ],
     "compressor": {
@@ -16,7 +16,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        1443,
+        1224,
         2
     ],
     "zarr_format": 2

diff --git a/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/coords/0/0 b/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/coords/0/0
diff --git a/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/offset0/.zarray b/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/offset0/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        16
+        15
     ],
     "compressor": {
         "blocksize": 0,
@@ -15,7 +15,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        16
+        15
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/offset0/0 b/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/offset0/0
diff --git a/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/offset1/.zarray b/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/offset1/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        16
+        15
     ],
     "compressor": {
         "blocksize": 0,
@@ -15,7 +15,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        16
+        15
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/offset1/0 b/bento/datasets/merfish_sample.zarr/shapes/cell_boundaries/offset1/0
diff --git a/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/Index/.zarray b/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/Index/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        15
+        12
     ],
     "compressor": {
         "blocksize": 0,
@@ -19,7 +19,7 @@
     ],
     "order": "C",
     "shape": [
-        15
+        12
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/Index/0 b/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/Index/0
diff --git a/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/coords/.zarray b/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/coords/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        2877,
+        2136,
         2
     ],
     "compressor": {
@@ -16,7 +16,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        2877,
+        2136,
         2
     ],
     "zarr_format": 2

diff --git a/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/coords/0/0 b/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/coords/0/0
diff --git a/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/offset0/.zarray b/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/offset0/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        16
+        13
     ],
     "compressor": {
         "blocksize": 0,
@@ -15,7 +15,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        16
+        13
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/offset0/0 b/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/offset0/0
diff --git a/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/offset1/.zarray b/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/offset1/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        16
+        13
     ],
     "compressor": {
         "blocksize": 0,
@@ -15,7 +15,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        16
+        13
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/offset1/0 b/bento/datasets/merfish_sample.zarr/shapes/nucleus_boundaries/offset1/0
diff --git a/bento/datasets/merfish_sample.zarr/table/table/X/.zattrs b/bento/datasets/merfish_sample.zarr/table/table/X/.zattrs
@@ -2,7 +2,7 @@
     "encoding-type": "csr_matrix",
     "encoding-version": "0.1.0",
     "shape": [
-        15,
+        14,
         135
     ]
 }
diff --git a/bento/datasets/merfish_sample.zarr/table/table/X/data/.zarray b/bento/datasets/merfish_sample.zarr/table/table/X/data/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        2025
+        1890
     ],
     "compressor": {
         "blocksize": 0,
@@ -15,7 +15,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        2025
+        1890
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/table/table/X/data/0 b/bento/datasets/merfish_sample.zarr/table/table/X/data/0
diff --git a/bento/datasets/merfish_sample.zarr/table/table/X/indices/.zarray b/bento/datasets/merfish_sample.zarr/table/table/X/indices/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        2025
+        1890
     ],
     "compressor": {
         "blocksize": 0,
@@ -15,7 +15,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        2025
+        1890
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/table/table/X/indices/0 b/bento/datasets/merfish_sample.zarr/table/table/X/indices/0
diff --git a/bento/datasets/merfish_sample.zarr/table/table/X/indptr/.zarray b/bento/datasets/merfish_sample.zarr/table/table/X/indptr/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        16
+        15
     ],
     "compressor": {
         "blocksize": 0,
@@ -15,7 +15,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        16
+        15
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/table/table/X/indptr/0 b/bento/datasets/merfish_sample.zarr/table/table/X/indptr/0
diff --git a/bento/datasets/merfish_sample.zarr/table/table/obs/_index/.zarray b/bento/datasets/merfish_sample.zarr/table/table/obs/_index/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        15
+        14
     ],
     "compressor": {
         "blocksize": 0,
@@ -19,7 +19,7 @@
     ],
     "order": "C",
     "shape": [
-        15
+        14
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/table/table/obs/_index/0 b/bento/datasets/merfish_sample.zarr/table/table/obs/_index/0
diff --git a/bento/datasets/merfish_sample.zarr/table/table/obs/cell_boundaries/.zarray b/bento/datasets/merfish_sample.zarr/table/table/obs/cell_boundaries/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        15
+        14
     ],
     "compressor": {
         "blocksize": 0,
@@ -19,7 +19,7 @@
     ],
     "order": "C",
     "shape": [
-        15
+        14
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/table/table/obs/cell_boundaries/0 b/bento/datasets/merfish_sample.zarr/table/table/obs/cell_boundaries/0
diff --git a/bento/datasets/merfish_sample.zarr/table/table/obs/region/codes/.zarray b/bento/datasets/merfish_sample.zarr/table/table/obs/region/codes/.zarray
@@ -1,6 +1,6 @@
 {
     "chunks": [
-        15
+        14
     ],
     "compressor": {
         "blocksize": 0,
@@ -15,7 +15,7 @@
     "filters": null,
     "order": "C",
     "shape": [
-        15
+        14
     ],
     "zarr_format": 2
 }
diff --git a/bento/datasets/merfish_sample.zarr/table/table/obs/region/codes/0 b/bento/datasets/merfish_sample.zarr/table/table/obs/region/codes/0