prepare - work around anndata bug (#1260)

* work around anndata bug 344

* fix accidental cut and paste error

* Use modified make_index_unique function

Temporarily copy code from https://github.com/theislab/anndata/pull/345
until the issue is resolved and released.

* Add notes and test for make_index_unique

* Lint fix

* Format python

Co-authored-by: Matt Weiden <538456+mweiden@users.noreply.github.com>

server/cli/launch.py

@@ -138,7 +138,7 @@ def dataset_args(func):
         "-t",
         default=DEFAULT_CONFIG.single_dataset__title,
         metavar="<text>",
-        help="Title to display. If omitted will use file name."
+        help="Title to display. If omitted will use file name.",
     )
     @click.option(
         "--about",
@@ -300,7 +300,7 @@ def launch(
     experimental_annotations_ontology_obo,
     experimental_enable_reembedding,
     config_file,
-    dump_default_config
+    dump_default_config,
 ):
     """Launch the cellxgene data viewer.
     This web app lets you explore single-cell expression data.
@@ -342,29 +342,22 @@ def launch(
             server__port=port,
             server__scripts=scripts,
             server__open_browser=open_browser,
-
             single_dataset__datapath=datapath,
             single_dataset__title=title,
             single_dataset__about=about,
             single_dataset__obs_names=obs_names,
             single_dataset__var_names=var_names,
-
             multi_dataset__dataroot=dataroot,
-
             user_annotations__enable=not disable_annotations,
             user_annotations__local_file_csv__file=annotations_file,
             user_annotations__local_file_csv__directory=annotations_dir,
             user_annotations__ontology__enable=experimental_annotations_ontology,
             user_annotations__ontology__obo_location=experimental_annotations_ontology_obo,
-
             presentation__max_categories=max_category_items,
-
             embeddings__names=embedding,
             embeddings__enable_reembedding=experimental_enable_reembedding,
-
             diffexp__enable=not disable_diffexp,
             diffexp__lfc_cutoff=diffexp_lfc_cutoff,
-
             adaptor__anndata_adaptor__backed=backed,
         )
 
@@ -386,6 +379,7 @@ def launch(
 
     # create the server
     from server.app.app import Server
+
     server = Server(matrix_data_cache_manager, user_annotations, app_config)
 
     if not app_config.server__verbose:
@@ -411,7 +405,8 @@ def launch(
             debug=app_config.server__debug,
             port=app_config.server__port,
             threaded=not app_config.server__debug,
-            use_debugger=False)
+            use_debugger=False,
+        )
     except OSError as e:
         if e.errno == errno.EADDRINUSE:
             raise click.ClickException("Port is in use, please specify an open port using the --port flag.") from e

server/cli/prepare.py

@@ -1,6 +1,7 @@
 from os.path import expanduser, isdir, isfile, sep, splitext
 
 import click
+import pandas as pd
 from numpy import ndarray, unique
 from scipy.sparse.csc import csc_matrix
 
@@ -23,12 +24,7 @@ from server.common.utils import sort_options
     show_default=True,
 )
 @click.option(
-    "--recipe",
-    "-r",
-    default="none",
-    type=click.Choice(["none", "seurat", "zheng17"]),
-    help="Preprocessing to run.",
-    show_default=True,
+    "--recipe", "-r", default="none", type=click.Choice(["none", "seurat", "zheng17"]), show_default=True,
 )
 @click.option("--output", "-o", default="", help="Save a new file to filename.", metavar="<filename>")
 @click.option("--plotting", "-p", default=False, is_flag=True, help="Generate plots.", show_default=True)
@@ -44,10 +40,16 @@ from server.common.utils import sort_options
     "(saved to adata.obs and adata.var; see scanpy.pp.calculate_qc_metrics for details).",
 )
 @click.option(
-    "--make-obs-names-unique", default=True, is_flag=True, help="Ensure obs index is unique.", show_default=True
+    "--make-obs-names-unique/--no-make-obs-names-unique",
+    default=True,
+    help="Ensure obs index is unique.",
+    show_default=True,
 )
 @click.option(
-    "--make-var-names-unique", default=True, is_flag=True, help="Ensure var index is unique.", show_default=True
+    "--make-var-names-unique/--no-make-var-names-unique",
+    default=True,
+    help="Ensure var index is unique.",
+    show_default=True,
 )
 @click.help_option("--help", "-h", help="Show this message and exit.")
 def prepare(
@@ -129,9 +131,9 @@ def prepare(
                 raise click.UsageError(f"var {set_var_names} not found, options are: {adata.var_keys()}")
             adata.var_names = adata.var[set_var_names]
         if make_obs_names_unique:
-            adata.obs_names_make_unique()
+            adata.obs.index = make_index_unique(adata.obs.index)
         if make_var_names_unique:
-            adata.var_names_make_unique()
+            adata.var.index = make_index_unique(adata.var.index)
         if not adata._obs.index.is_unique:
             click.echo("Warning: obs index is not unique")
         if not adata._var.index.is_unique:
@@ -221,3 +223,51 @@ def prepare(
         adata.write(output)
 
     click.echo("[cellxgene] Success!")
+
+
+# TODO (mweiden): remove this once this issue is resolved https://github.com/theislab/anndata/issues/344
+# Note: tentative solution here https://github.com/theislab/anndata/pull/345
+def make_index_unique(index: pd.Index, join: str = "-"):
+    """
+    Makes the index unique by appending a number string to each duplicate index element: '1', '2', etc.
+
+    If a tentative name created by the algorithm already exists in the index, it tries the next integer in the sequence.
+
+    The first occurrence of a non-unique value is ignored.
+    Parameters
+    ----------
+    join
+         The connecting string between name and integer.
+    Examples
+    --------
+    >>> from anndata import AnnData
+    >>> adata1 = AnnData(np.ones((3, 2)), dict(obs_names=['a', 'b', 'c']))
+    >>> adata2 = AnnData(np.zeros((3, 2)), dict(obs_names=['d', 'b', 'b']))
+    >>> adata = adata1.concatenate(adata2)
+    >>> adata.obs_names
+    Index(['a', 'b', 'c', 'd', 'b', 'b'], dtype='object')
+    >>> adata.obs_names_make_unique()
+    >>> adata.obs_names
+    Index(['a', 'b', 'c', 'd', 'b-1', 'b-2'], dtype='object')
+    """
+    if index.is_unique:
+        return index
+    from collections import defaultdict
+
+    values = index.values
+    values_set = set(values)
+    indices_dup = index.duplicated(keep="first")
+    values_dup = values[indices_dup]
+    counter = defaultdict(lambda: 0)
+    for i, v in enumerate(values_dup):
+        while True:
+            counter[v] += 1
+            tentative_new_name = v + join + str(counter[v])
+            if tentative_new_name not in values_set:
+                values_set.add(tentative_new_name)
+                values_dup[i] = tentative_new_name
+                break
+
+    values[indices_dup] = values_dup
+    index = pd.Index(values)
+    return index

server/common/app_config.py

@@ -344,6 +344,7 @@ class AppConfig(object):
         # cxg
         self.__check_attr("adaptor__cxg_adaptor__tiledb_ctx", dict)
         from server.data_cxg.cxg_adaptor import CxgAdaptor
+
         CxgAdaptor.set_tiledb_context(self.adaptor__cxg_adaptor__tiledb_ctx)
 
         # anndata

server/data_cxg/cxg_adaptor.py

@@ -17,11 +17,9 @@ import threading
 class CxgAdaptor(DataAdaptor):
 
     # TODO:  The tiledb context parameters should be a configuration option
-    tiledb_ctx = tiledb.Ctx({
-        "sm.tile_cache_size": 8 * 1024 * 1024 * 1024,
-        "sm.num_reader_threads": 32,
-        "vfs.s3.region": "us-east-1"
-    })
+    tiledb_ctx = tiledb.Ctx(
+        {"sm.tile_cache_size": 8 * 1024 * 1024 * 1024, "sm.num_reader_threads": 32, "vfs.s3.region": "us-east-1"}
+    )
 
     def __init__(self, data_locator, config=None):
         super().__init__(config)

server/eb/app.py

@@ -38,9 +38,7 @@ try:
 
     if dataroot:
         logging.info(f"Configuration from CXG_DATAROOT")
-        app_config.update(
-            multi_dataset__dataroot=dataroot,
-        )
+        app_config.update(multi_dataset__dataroot=dataroot,)
 
     # features are unsupported in the current hosted server
     app_config.update(

server/test/test_anndata_adaptor.py

@@ -40,7 +40,7 @@ class AdaptorTest(unittest.TestCase):
             "single_dataset__var_names": None,
             "diffexp__lfc_cutoff": 0.01,
             "adaptor__anndata_adaptor__backed": self.backed,
-            "single_dataset__datapath" : self.data_locator
+            "single_dataset__datapath": self.data_locator,
         }
         config = AppConfig()
         config.update(**args)

server/test/test_cli_prepare.py

@@ -0,0 +1,15 @@
+import unittest
+
+import pandas as pd
+
+from server.cli.prepare import make_index_unique
+
+
+class CLIPrepareTests(unittest.TestCase):
+    """ Test cases for CLI prepare logic """
+
+    def test_make_index_unique(self):
+        index = pd.Index(["SNORD113", "SNORD113", "SNORD113-1"])
+        result = make_index_unique(index)
+        expected = pd.Index(["SNORD113", "SNORD113-2", "SNORD113-1"])
+        self.assertTrue(all(left == right for left, right in zip(result.values, expected.values)))