htsworkflow/submission/test/test_encoded.py

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