cellmaps-utils 0.1.0a22__tar.gz → 0.2.0__tar.gz

cellmaps_utils-0.2.0/HISTORY.rst

@@ -0,0 +1,22 @@
+=======
+History
+=======
+
+0.2.0 (2024-02-20)
+------------------
+
+* Bumped fairscape-cli dependency to ``0.1.14`` to support schemas
+
+* Added support for ``schema`` to **data_dict** parameter in ``ProvenanceUtil.register_dataset()``
+
+* Added ``--release`` flag to ``cellmaps_utilscmd.py rocratetable`` and
+  in output table renamed "Name of Computation" to "Name" as well as
+  added "Type", "Cell Line", "Treatment", "Gene set", and "Version" to
+  table output
+
+* Set default logging level to ``ERROR`` for ``cellmaps_utilscmd.py`` command
+
+0.1.0 (2024-01-01)
+------------------
+
+* First release on PyPI.

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cellmaps_utils
-Version: 0.1.0a22
+Version: 0.2.0
 Summary: Contains utilities needed by Cell Maps tools
 Home-page: https://github.com/idekerlab/cellmaps_utils
 Author: Cell Maps team
@@ -21,6 +21,10 @@ Description: =============================
                 :target: https://cellmaps-utils.readthedocs.io/en/latest/?badge=latest
                 :alt: Documentation Status
         
+        .. image:: https://zenodo.org/badge/599754117.svg
+                :target: https://zenodo.org/doi/10.5281/zenodo.10607322
+                :alt: Zenodo DOI badge
+        
         Contains utilities needed by various tools in the cell maps toolkit. 
         
         
@@ -116,7 +120,21 @@ Description: =============================
         History
         =======
         
-        0.1.0 (2023-05-01)
+        0.2.0 (2024-02-20)
+        ------------------
+        
+        * Bumped fairscape-cli dependency to ``0.1.14`` to support schemas
+        
+        * Added support for ``schema`` to **data_dict** parameter in ``ProvenanceUtil.register_dataset()``
+        
+        * Added ``--release`` flag to ``cellmaps_utilscmd.py rocratetable`` and
+          in output table renamed "Name of Computation" to "Name" as well as
+          added "Type", "Cell Line", "Treatment", "Gene set", and "Version" to
+          table output
+        
+        * Set default logging level to ``ERROR`` for ``cellmaps_utilscmd.py`` command
+        
+        0.1.0 (2024-01-01)
         ------------------
         
         * First release on PyPI.

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/README.rst

@@ -13,6 +13,10 @@ Cell Maps Pipeline Utilities
         :target: https://cellmaps-utils.readthedocs.io/en/latest/?badge=latest
         :alt: Documentation Status
 
+.. image:: https://zenodo.org/badge/599754117.svg
+        :target: https://zenodo.org/doi/10.5281/zenodo.10607322
+        :alt: Zenodo DOI badge
+
 Contains utilities needed by various tools in the cell maps toolkit. 
 
 

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/cellmaps_utils/__init__.py

@@ -4,6 +4,6 @@
 
 __author__ = 'Cell Maps team'
 __email__ = 'tools@cm4ai.org'
-__version__ = '0.1.0a22'
+__version__ = '0.2.0'
 __repo_url__ = 'https://github.com/idekerlab/cellmaps_utils'
 __description__ = 'Contains utilities needed by Cell Maps tools'

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/cellmaps_utils/cellmaps_utilscmd.py

@@ -54,12 +54,12 @@ def _parse_arguments(desc, args):
                              'logging.config.html#logging-config-fileformat '
                              'Setting this overrides -v parameter which uses '
                              ' default logger. (default None)')
-    parser.add_argument('--verbose', '-v', action='count', default=0,
+    parser.add_argument('--verbose', '-v', action='count', default=1,
                         help='Increases verbosity of logger to standard '
                              'error for log messages in this module. Messages are '
                              'output at these python logging levels '
-                             '-v = ERROR, -vv = WARNING, -vvv = INFO, '
-                             '-vvvv = DEBUG, -vvvvv = NOTSET (default no '
+                             '-v = WARNING, -vv = INFO, '
+                             '-vvv = DEBUG, -vvvv = NOTSET (default ERROR '
                              'logging)')
     parser.add_argument('--version', action='version',
                         version=('%(prog)s ' +

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/cellmaps_utils/provenance.py

@@ -669,8 +669,9 @@ class ProvenanceUtil(object):
                 cmd.append(entry)
         if used_dataset is not None:
             for entry in used_dataset:
-                cmd.append('--used-dataset')
-                cmd.append(entry)
+                if entry is not None:
+                    cmd.append('--used-dataset')
+                    cmd.append(entry)
 
         if generated is not None:
             for entry in generated:
@@ -703,11 +704,6 @@ class ProvenanceUtil(object):
         ``ro-crate-metadata.json`` file stored in **rocrate_path**
         directory.
 
-        .. warning::
-
-            `fairscape-cli <https://github.com/fairscape>`__ 0.1.5 always fails this call
-            `See Issue #7 <https://github.com/fairscape/fairscape-cli/issues/7>`__
-
         :param name: Name of software
         :type name: str
         :param description: Description of software
@@ -791,14 +787,12 @@ class ProvenanceUtil(object):
              'date-published': 'Date dataset was published MM-DD-YYYY',
              'description': 'Description of dataset',
              'data-format': 'Format of data',
+             'schema': Path or URL to schema file in JSON format
              'keywords': ['keyword1','keyword2']}
 
-        .. warning::
+        .. versionchanged:: 0.2.0
 
-            `fairscape-cli <https://github.com/fairscape>`__ 0.1.5 fails when
-            skipping copy is ``True`` or if performing copy where
-            **source_file** is already in **rocrate_path** directory.
-            `See Issue #6 <https://github.com/fairscape/fairscape-cli/issues/6>`__
+            Added support for ``schema`` in **data_dict** passed in
 
 
         :param rocrate_path: Path to directory with registered rocrate
@@ -839,6 +833,9 @@ class ProvenanceUtil(object):
         else:
             cmd.extend(self._get_keywords(keywords=data_dict['keywords']))
 
+        if 'schema' in data_dict:
+            cmd.extend(['--schema', data_dict['schema']])
+
         if guid is None:
             guid = self._generate_guid(data_type='dataset',
                                        rocrate_path=rocrate_path)

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/cellmaps_utils/tabletool.py

@@ -23,7 +23,7 @@ class TableFromROCrates(BaseCommandLineTool):
 
     ID_COL = 'FAIRSCAPE ARK ID'
     DATE_COL = 'Date'
-    COMPUTATION_COL = 'Name of Computation'
+    COMPUTATION_COL = 'Name'
     DESCRIPTION_COL = 'Description'
     KEYWORDS_COL = 'Keywords'
     DOWNLOAD_COL = 'Download RO-Crate Data Package'
@@ -31,8 +31,24 @@ class TableFromROCrates(BaseCommandLineTool):
     GENERATED_COL = 'Generated By Software'
     OUTPUT_COL = 'Output Dataset'
     RESPONSIBLE_COL = 'Responsible Lab'
+    TYPE_COL = 'Type'
+    CELL_LINE_COL = 'Cell Line'
+    TREATMENT_COL = 'Treatment'
+    GENESET_COL = 'Gene set'
+    VERSION_COL = 'Version'
+
+    DATA_ROCRATE = 'Data'
+    MODEL_ROCRATE = 'Model'
+    INTERMEDIATE_ROCRATE = 'Intermediate'
+    OTHER_ROCRATE = 'Other'
 
     COLUMNS = [ID_COL, DATE_COL,
+               VERSION_COL,
+               TYPE_COL,
+               CELL_LINE_COL,
+               TREATMENT_COL,
+               GENESET_COL,
+               GENERATED_COL,
                COMPUTATION_COL,
                DESCRIPTION_COL, KEYWORDS_COL,
                DOWNLOAD_COL,
@@ -54,10 +70,70 @@ class TableFromROCrates(BaseCommandLineTool):
         self._outdir = os.path.abspath(theargs.outdir)
         self._rocrates = theargs.rocrates
         self._date = theargs.date
+        self._version = theargs.release
         self._provenance_utils = provenance_utils
         self._downloadurlprefix = theargs.downloadurlprefix
         self._input_data_dict = theargs.__dict__
 
+    def _get_rocrate_type(self, compname):
+        """
+        Gets type of rocrate, which can be Data, Model, or Intermediate
+
+        :return:
+        :rtype: str
+        """
+        if compname in ['IF images', 'AP-MS', 'CRISPR']:
+            return TableFromROCrates.DATA_ROCRATE
+        if compname in ['Hierarchy']:
+            return TableFromROCrates.MODEL_ROCRATE
+        return TableFromROCrates.INTERMEDIATE_ROCRATE
+
+    def _get_cell_line(self, keywords):
+        """
+        Looks for cell line in keywords
+
+        :param keywords:
+        :type keywords: list
+        :return:
+        :rtype: str
+        """
+        if 'MDA-MB-468' in keywords:
+            return 'MDA-MB-468'
+        return 'Unknown'
+
+    def _get_treatment(self, keywords):
+        """
+        Gets treatment from keywords
+
+        :param keywords:
+        :return:
+        """
+        treatments = []
+        if 'untreated' in keywords:
+            treatments.append('untreated')
+        if 'vorinostat' in keywords:
+            treatments.append('vorinostat')
+        if 'paclitaxel' in keywords:
+            treatments.append('paclitaxel')
+        return ','.join(treatments)
+
+    def _get_geneset(self, keywords):
+        """
+        Gets treatment from keywords
+
+        :param keywords:
+        :return:
+        """
+        genesets = []
+
+        if 'chromatin' in keywords:
+            genesets.append('chromatin')
+        if 'metabolic' in keywords:
+            genesets.append('metabolic')
+        if len(genesets) == 0:
+            return 'Unknown'
+        return ','.join(genesets)
+
     def run(self):
         """
 
@@ -95,6 +171,11 @@ class TableFromROCrates(BaseCommandLineTool):
                 logger.debug(rocrate + ': ' + 'Keywords: ' + str(prov_attrs.get_keywords()))
                 row = {TableFromROCrates.ID_COL: self._provenance_utils.get_id_of_rocrate(rocrate=rocrate_dict),
                        TableFromROCrates.DATE_COL: self._date,
+                       TableFromROCrates.VERSION_COL: self._version,
+                       TableFromROCrates.TYPE_COL: self._get_rocrate_type(comp_name),
+                       TableFromROCrates.CELL_LINE_COL: self._get_cell_line(prov_attrs.get_keywords()),
+                       TableFromROCrates.TREATMENT_COL: self._get_treatment(prov_attrs.get_keywords()),
+                       TableFromROCrates.GENESET_COL: self._get_geneset(prov_attrs.get_keywords()),
                        TableFromROCrates.COMPUTATION_COL: comp_name,
                        TableFromROCrates.DESCRIPTION_COL: prov_attrs.get_description(),
                        TableFromROCrates.KEYWORDS_COL: ','.join(prov_attrs.get_keywords()),
@@ -278,6 +359,8 @@ class TableFromROCrates(BaseCommandLineTool):
                             help='Path to RO-Crates used for table generation. ')
         parser.add_argument('--date', default=date.today().strftime('%Y-%m-%d'),
                             help='Date to list in table for RO-Crates')
+        parser.add_argument('--release', default='0.1 alpha',
+                            help='Version of release')
         parser.add_argument('--downloadurlprefix', default='https://g-9b3b6e.9ad93.a567.data.globus.org/Data/cm4ai_0.1alpha/')
 
         return parser

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/cellmaps_utils.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cellmaps-utils
-Version: 0.1.0a22
+Version: 0.2.0
 Summary: Contains utilities needed by Cell Maps tools
 Home-page: https://github.com/idekerlab/cellmaps_utils
 Author: Cell Maps team
@@ -21,6 +21,10 @@ Description: =============================
                 :target: https://cellmaps-utils.readthedocs.io/en/latest/?badge=latest
                 :alt: Documentation Status
         
+        .. image:: https://zenodo.org/badge/599754117.svg
+                :target: https://zenodo.org/doi/10.5281/zenodo.10607322
+                :alt: Zenodo DOI badge
+        
         Contains utilities needed by various tools in the cell maps toolkit. 
         
         
@@ -116,7 +120,21 @@ Description: =============================
         History
         =======
         
-        0.1.0 (2023-05-01)
+        0.2.0 (2024-02-20)
+        ------------------
+        
+        * Bumped fairscape-cli dependency to ``0.1.14`` to support schemas
+        
+        * Added support for ``schema`` to **data_dict** parameter in ``ProvenanceUtil.register_dataset()``
+        
+        * Added ``--release`` flag to ``cellmaps_utilscmd.py rocratetable`` and
+          in output table renamed "Name of Computation" to "Name" as well as
+          added "Type", "Cell Line", "Treatment", "Gene set", and "Version" to
+          table output
+        
+        * Set default logging level to ``ERROR`` for ``cellmaps_utilscmd.py`` command
+        
+        0.1.0 (2024-01-01)
         ------------------
         
         * First release on PyPI.

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/cellmaps_utils.egg-info/SOURCES.txt

@@ -45,6 +45,8 @@ docs/_build/html/_static/file.png
 docs/_build/html/_static/minus.png
 docs/_build/html/_static/plus.png
 tests/__init__.py
+tests/test_basecommandlinetool.py
+tests/test_helloworldcommand.py
 tests/test_integation_provenance.py
 tests/test_logutils.py
 tests/test_provenance.py

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/cellmaps_utils.egg-info/requires.txt

@@ -1,4 +1,4 @@
-fairscape-cli==0.1.8
+fairscape-cli==0.1.14
 scipy
 scikit-learn
 pandas

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/docs/index.rst

@@ -13,6 +13,10 @@ Cell Maps Utilities
         :target: https://cellmaps-utils.readthedocs.io/en/latest/?badge=latest
         :alt: Documentation Status
 
+.. image:: https://zenodo.org/badge/599754117.svg
+        :target: https://zenodo.org/doi/10.5281/zenodo.10607322
+        :alt: Zenodo DOI badge
+
 Contains utilities needed by various tools in the cell maps toolkit. 
 
 .. toctree::

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/setup.py

@@ -28,7 +28,7 @@ with open('HISTORY.rst') as history_file:
     history = history_file.read()
 
 
-requirements = ['fairscape-cli==0.1.8',
+requirements = ['fairscape-cli==0.1.14',
                 'scipy',
                 'scikit-learn',
                 'pandas',

cellmaps_utils-0.2.0/tests/test_basecommandlinetool.py

@@ -0,0 +1,38 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+"""Tests for `BaseCommandLineTool`."""
+
+import os
+import argparse
+import shutil
+import tempfile
+import unittest
+
+from cellmaps_utils.exceptions import CellMapsError
+from cellmaps_utils.basecmdtool import BaseCommandLineTool
+
+
+class TestBaseCommandLineTool(unittest.TestCase):
+    """Tests for `BaseCommandLineTool`"""
+
+    def setUp(self):
+        """Set up test fixtures, if any."""
+
+    def tearDown(self):
+        """Tear down test fixtures, if any."""
+
+    def test_run(self):
+        cmdtool = BaseCommandLineTool()
+        try:
+            cmdtool.run()
+            self.fail('Expected exception')
+        except CellMapsError as ce:
+            self.assertTrue('Must be implemented by subclass' in str(ce))
+
+    def test_add_subparser(self):
+        try:
+            BaseCommandLineTool.add_subparser(None)
+            self.fail('Expected exception')
+        except CellMapsError as ce:
+            self.assertTrue('Must be implemented by subclass' in str(ce))

cellmaps_utils-0.2.0/tests/test_helloworldcommand.py

@@ -0,0 +1,35 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+"""Tests for `HelloWorldCommand`."""
+
+import os
+import argparse
+import shutil
+import tempfile
+import unittest
+from unittest.mock import MagicMock
+
+from cellmaps_utils.exceptions import CellMapsError
+from cellmaps_utils.basecmdtool import HelloWorldCommand
+
+
+class TestHelloWorldCommand(unittest.TestCase):
+    """Tests for `HelloWorldCommand`"""
+
+    def setUp(self):
+        """Set up test fixtures, if any."""
+
+    def tearDown(self):
+        """Tear down test fixtures, if any."""
+
+    def test_run(self):
+
+        cmdtool = HelloWorldCommand(MagicMock())
+        self.assertEqual(0, cmdtool.run())
+
+    def test_add_subparser(self):
+        subparsers = MagicMock()
+        subparsers.add_parser(return_value=True)
+        self.assertTrue(HelloWorldCommand.add_subparser(subparsers))
+

{cellmaps_utils-0.1.0a22 → cellmaps_utils-0.2.0}/tests/test_provenance.py

@@ -282,6 +282,32 @@ class TestProvenanceUtil(unittest.TestCase):
         finally:
             shutil.rmtree(temp_dir)
 
+    def test_register_dataset_with_schema(self):
+        temp_dir = tempfile.mkdtemp()
+        try:
+            subdir = os.path.join(temp_dir, 'input')
+            os.makedirs(subdir, mode=0o755)
+            src_file = os.path.join(subdir, 'xx')
+            with open(src_file, 'w') as f:
+                f.write('hi')
+
+            prov = ProvenanceUtil()
+            prov.register_rocrate(temp_dir, name='some 10 character name', description='10 character description')
+            d_id = prov.register_dataset(temp_dir,
+                                         source_file=src_file,
+                                         skip_copy=False,
+                                         data_dict={'name': 'Name of dataset',
+                                                    'author': 'Author of dataset',
+                                                    'version': 'Version of dataset',
+                                                    'date-published': 'Date dataset was published MM-DD-YYYY',
+                                                    'description': 'Description of dataset',
+                                                    'schema': 'https://foo.com',
+                                                    'data-format': 'Format of data'})
+            self.assertIsNotNone(d_id)
+
+        finally:
+            shutil.rmtree(temp_dir)
+
     def test_register_dataset_with_keywords(self):
 
         temp_dir = tempfile.mkdtemp()

cellmaps_utils-0.1.0a22/HISTORY.rst

@@ -1,8 +0,0 @@
-=======
-History
-=======
-
-0.1.0 (2023-05-01)
-------------------
-
-* First release on PyPI.