htsworkflow/submission/test/test_encoded.py

   1 from __future__ import absolute_import, print_function
   2
   3 import json
   4 import os
   5 from pprint import pprint
   6 from unittest import TestCase, TestSuite, defaultTestLoader, skip
   7
   8 from htsworkflow.submission.encoded import (ENCODED,
   9      ENCODED_CONTEXT,
  10      ENCODED_NAMESPACES
  11 )
  12
  13 class TestEncoded(TestCase):
  14     def test_prepare_url(self):
  15         encode = ENCODED('test.encodedcc.edu')
  16
  17         tests = [
  18             ('/experiments', 'https://test.encodedcc.edu/experiments'),
  19             ('/experiments/ENCLB045ZZZ',
  20              'https://test.encodedcc.edu/experiments/ENCLB045ZZZ'),
  21             ('https://www.encodedcc.edu/experiments/ENCLB045ZZZ',
  22              'https://www.encodedcc.edu/experiments/ENCLB045ZZZ'),
  23         ]
  24         for url, result in tests:
  25             self.assertEqual(encode.prepare_url(url), result)
  26
  27     def test_validate(self):
  28         """Test validation
  29         """
  30         schema_file = os.path.join(os.path.dirname(__file__), 'library.json')
  31         schema = json.loads(open(schema_file, 'r').read())
  32
  33         obj = {u'@id': u'/libraries/ENCLB045ZZZ/',
  34                u'@type': [u'library', u'item'],
  35                u'accession': u'ENCLB045ZZZ',
  36                u'aliases': [],
  37                u'alternate_accessions': [],
  38                u'award': u'/awards/U54HG006998/',
  39                u'biosample': u'/biosamples/ENCBS089RNA/',
  40                u'date_created': u'2014-01-14T19:44:51.061770+00:00',
  41                u'depleted_in_term_id': [],
  42                u'depleted_in_term_name': [],
  43                u'documents': [],
  44                u'extraction_method': u'Ambion mirVana',
  45                u'fragmentation_method': u'Illumina/Nextera tagmentation',
  46                u'lab': u'/labs/barbara-wold/',
  47                u'library_size_selection_method': u'SPRI beads',
  48                u'lysis_method': u'Ambion mirVana',
  49                u'nucleic_acid_term_id': u'SO:0000871',
  50                u'nucleic_acid_term_name': u'polyadenylated mRNA',
  51                u'paired_ended': False,
  52                u'schema_version': u'2',
  53                u'size_range': u'>200',
  54                u'status': u'CURRENT',
  55                u'strand_specificity': False,
  56                u'submitted_by': u'/users/0e3dde9b-aaf9-42dd-87f7-975a85072ed2/',
  57                u'treatments': [],
  58                u'uuid': u'42c46028-708f-4347-a3df-2c82dfb021c4'}
  59         encode = ENCODED('www.encodeproject.org')
  60         encode.schemas[u'library'] = schema
  61         encode.validate(obj)
  62         self.assertTrue('@id' in obj)
  63
  64     def test_create_context(self):
  65         linked_id = {'@type': '@id'}
  66         library = { '@id': '/libraries/1234', '@type': ['library', 'item'] }
  67
  68         encode = ENCODED('test.encodedcc.org')
  69         url = encode.prepare_url(library['@id'])
  70         context = encode.create_jsonld_context(library, url)
  71         self.assertEqual(context['@vocab'], 'https://test.encodedcc.org/profiles/library.json#')
  72         self.assertEqual(context['award'], linked_id )
  73         self._verify_context(context, 'library')
  74         # namespaces not added yet.
  75         self.assertRaises(AssertionError, self._verify_namespaces, context)
  76         encode.add_jsonld_namespaces(context)
  77         self._verify_namespaces(context)
  78
  79     def test_add_context(self):
  80         """Checking to make sure nested @base and @vocab urls are set correctly
  81         """
  82         obj = {
  83             "nucleic_acid_term_name": "RNA",
  84             "accession": "ENCLB044ZZZ",
  85             "@id": "/libraries/ENCLB044ZZZ/",
  86             "schema_version": "1",
  87             "@type": [
  88                 "library",
  89                 "item"
  90             ],
  91             "lysis_method": "Ambion mirVana",
  92             "nucleic_acid_term_id": "SO:0000356",
  93             "biosample": {
  94                 "biosample_term_name": "GM12878",
  95                 "description": "B-lymphocyte, lymphoblastoid, International HapMap Project - CEPH/Utah - European Caucasion, Epstein-Barr Virus",
  96                 "accession": "ENCBS090RNA",
  97                 "date_created": "2013-10-29T21:15:29.144260+00:00",
  98                 "@id": "/biosamples/ENCBS090RNA/",
  99                 "aliases": [
 100                 "brenton-graveley:GM12878-2",
 101                 "thomas-gingeras:191WC"
 102                 ],
 103                 "organism": "/organisms/human/",
 104                 "@type": [
 105                 "biosample",
 106                 "item"
 107                 ]
 108             },
 109         }
 110
 111         encode = ENCODED('test.encodedcc.org')
 112         bio_base = encode.prepare_url(obj['biosample']['@id'])
 113
 114         url = encode.prepare_url('/libraries/ENCLB044ZZZ/?format=json&embed=False')
 115         obj_type = encode.get_object_type(obj)
 116         schema_url = encode.get_schema_url(obj_type)
 117         encode.add_jsonld_context(obj, url)
 118
 119         self.assertEqual(obj['biosample']['@context']['@base'], bio_base)
 120         self.assertEqual(obj['@context']['@vocab'], schema_url)
 121         self._verify_context(obj['@context'], 'library')
 122         self._verify_namespaces(obj['@context'])
 123         self._verify_context(obj['biosample']['@context'], 'biosample')
 124         self.assertEqual(obj['@context']['rdf'], 'http://www.w3.org/1999/02/22-rdf-syntax-ns#')
 125         self.assertEqual(obj['@context']['OBO'], 'http://purl.obolibrary.org/obo/')
 126
 127
 128     def test_convert_search_to_jsonld(self):
 129         example = {'count': {'biosamples': 2},
 130                    'portal_title': 'ENCODE',
 131                    'title': 'Search',
 132                    'notification': 'Success',
 133                    'filters': [],
 134                    '@id': '/search/?searchTerm=wold',
 135                    '@type': ['search'],
 136                    'facets': [],
 137                     '@graph': [{
 138                     u'@id': u'/biosamples/ENCBS125ENC/',
 139                     u'@type': [u'biosample', u'item'],
 140                     u'accession': u'ENCBS125ENC',
 141                     u'award.rfa': u'ENCODE2-Mouse',
 142                     u'biosample_term_name': u'myocyte',
 143                     u'biosample_type': u'in vitro differentiated cells',
 144                     u'characterizations.length': [],
 145                     u'constructs.length': [],
 146                     u'lab.title': u'Barbara Wold, Caltech',
 147                     u'life_stage': u'unknown',
 148                     u'organism.name': u'mouse',
 149                     u'source.title': u'Barbara Wold',
 150                     u'status': u'CURRENT',
 151                     u'treatments.length': []},
 152                     {u'@id': u'/biosamples/ENCBS126ENC/',
 153                     u'@type': [u'biosample', u'item'],
 154                     u'accession': u'ENCBS126ENC',
 155                     u'award.rfa': u'ENCODE2-Mouse',
 156                     u'biosample_term_name': u'myocyte',
 157                     u'biosample_type': u'in vitro differentiated cells',
 158                     u'characterizations.length': [],
 159                     u'constructs.length': [],
 160                     u'lab.title': u'Barbara Wold, Caltech',
 161                     u'life_stage': u'unknown',
 162                     u'organism.name': u'mouse',
 163                     u'source.title': u'Barbara Wold',
 164                     u'status': u'CURRENT',
 165                     u'treatments.length': []},
 166                     ]}
 167
 168         encode = ENCODED('test.encodedcc.org')
 169         result = encode.convert_search_to_jsonld(example)
 170         for obj in result['@graph']:
 171             self.assertNotIn('award.rfa', obj)
 172
 173     def _verify_context(self, context, obj_type):
 174         for context_key in [None, obj_type]:
 175             for k in ENCODED_CONTEXT[context_key]:
 176                 self.assertIn(k, context)
 177                 self.assertEqual(ENCODED_CONTEXT[context_key][k], context[k])
 178
 179     def _verify_namespaces(self, context):
 180         for k in ENCODED_NAMESPACES:
 181             self.assertIn(k, context)
 182             self.assertEqual(ENCODED_NAMESPACES[k], context[k])
 183
 184 def suite():
 185     suite = TestSuite()
 186     suite.addTests(
 187         defaultTestLoader.loadTestsFromTestCase(TestEncoded))
 188     return suite
 189
 190 if __name__ == "__main__":
 191     from unittest import main
 192     main(defaultTest='suite')