htsworkflow/submission/test/test_daf.py

   1 from StringIO import StringIO
   2 import unittest
   3
   4 from htsworkflow.submission import daf
   5 from htsworkflow.util.rdfhelp import \
   6      dafTermOntology, \
   7      rdfNS, \
   8      submissionLog, \
   9      submissionOntology, \
  10      get_model, \
  11      get_serializer
  12
  13 import RDF
  14
  15 test_daf = """# Lab and general info
  16 grant             Hardison
  17 lab               Caltech-m
  18 dataType          ChipSeq
  19 variables         cell, antibody,sex,age,strain,control
  20 compositeSuffix   CaltechHistone
  21 assembly          mm9
  22 dafVersion        2.0
  23 validationSettings validateFiles.bam:mismatches=2,bamPercent=99.9;validateFiles.fastq:quick=1000
  24
  25 # Track/view definition
  26 view             FastqRd1
  27 longLabelPrefix  Caltech Fastq Read 1
  28 type             fastq
  29 hasReplicates    yes
  30 required         no
  31
  32 view             Signal
  33 longLabelPrefix  Caltech Histone Signal
  34 type             bigWig
  35 hasReplicates    yes
  36 required         no
  37 """
  38
  39 class TestDAF(unittest.TestCase):
  40     def test_parse(self):
  41
  42         parsed = daf.fromstring(test_daf)
  43
  44         self.failUnlessEqual(parsed['assembly'], 'mm9')
  45         self.failUnlessEqual(parsed['grant'], 'Hardison')
  46         self.failUnlessEqual(len(parsed['variables']), 6)
  47         self.failUnlessEqual(len(parsed['views']), 2)
  48         self.failUnlessEqual(len(parsed['views']['FastqRd1']), 5)
  49         self.failUnlessEqual(len(parsed['views']['Signal']), 5)
  50         signal = parsed['views']['Signal']
  51         self.failUnlessEqual(signal['required'], False)
  52         self.failUnlessEqual(signal['longLabelPrefix'],
  53                              'Caltech Histone Signal')
  54
  55     def test_rdf(self):
  56
  57         parsed = daf.fromstring(test_daf)
  58         #mem = RDF.Storage(storage_name='hashes',
  59         #                  options_string='hash-type="memory"'),
  60         mem = RDF.MemoryStorage()
  61         model = RDF.Model(mem)
  62
  63         name = 'cursub'
  64         subNS = RDF.NS(str(submissionLog[name].uri))
  65         daf.add_to_model(model, parsed, name)
  66
  67         signal_view_node = RDF.Node(subNS['/view/Signal'].uri)
  68         writer = get_serializer()
  69         turtle =  writer.serialize_model_to_string(model)
  70         #print turtle
  71
  72         self.failUnless(str(signal_view_node) in turtle)
  73
  74         statements = list(model.find_statements(
  75             RDF.Statement(
  76                 signal_view_node, None, None)))
  77         self.failUnlessEqual(len(statements), 5)
  78
  79
  80 def dump_model(model):
  81     writer = get_serializer()
  82     turtle =  writer.serialize_model_to_string(model)
  83     print turtle
  84
  85 class TestDAFMapper(unittest.TestCase):
  86     def test_create_mapper_add_pattern(self):
  87         name = 'testsub'
  88         test_daf_stream = StringIO(test_daf)
  89         mapper = daf.DAFMapper(name, daf_file=test_daf_stream)
  90         pattern = '.bam\Z(?ms)'
  91         mapper.add_pattern('Signal', pattern)
  92
  93         s = RDF.Statement(daf.get_view_namespace(name)['Signal'],
  94                           dafTermOntology['filename_re'],
  95                           None)
  96         search = list(mapper.model.find_statements(s))
  97         self.failUnlessEqual(len(search), 1)
  98         self.failUnlessEqual(str(search[0].subject),
  99                              str(submissionLog['testsub/view/Signal']))
 100         self.failUnlessEqual(str(search[0].predicate),
 101                              str(dafTermOntology['filename_re']))
 102         #self.failUnlessEqual(search[0].object.literal_value['string'], pattern)
 103
 104     def test_find_one_view(self):
 105         model = get_model()
 106
 107         parser = RDF.Parser(name='turtle')
 108         parser.parse_string_into_model(model, '''
 109 @prefix dafTerm:<http://jumpgate.caltech.edu/wiki/UcscDaf#> .
 110
 111 <%(submissionLog)s/testfind/view/Signal> dafTerm:filename_re ".*\\\\.bam" .
 112 <%(submissionLog)s/testfind/view/FastqRd1> dafTerm:filename_re ".*_r1\\\\.fastq" .
 113 ''' % {'submissionLog': 'http://jumpgate.caltech.edu/wiki/SubmissionsLog'},
 114         'http://blank')
 115         name = 'testfind'
 116         test_stream = StringIO(test_daf)
 117         daf_mapper = daf.DAFMapper(name, daf_file=test_stream, model=model)
 118
 119         view = daf_mapper.find_view('filename_r1.fastq')
 120         self.failUnlessEqual(str(view),
 121                              str(submissionLog['testfind/view/FastqRd1']))
 122
 123         #writer = get_serializer()
 124         #turtle =  writer.serialize_model_to_string(model)
 125         #print turtle
 126
 127     def test_find_overlapping_view(self):
 128         model = get_model()
 129
 130         parser = RDF.Parser(name='turtle')
 131         parser.parse_string_into_model(model, '''
 132 @prefix dafTerm:<http://jumpgate.caltech.edu/wiki/UcscDaf#> .
 133
 134 <%(submissionLog)s/testfind/view/fastq> dafTerm:filename_re ".*\\\\.fastq" .
 135 <%(submissionLog)s/testfind/view/FastqRd1> dafTerm:filename_re ".*_r1\\\\.fastq" .
 136 ''' % {'submissionLog': 'http://jumpgate.caltech.edu/wiki/SubmissionsLog'},
 137         'http://blank')
 138         name = 'testfind'
 139         test_stream = StringIO(test_daf)
 140         daf_mapper = daf.DAFMapper(name, daf_file=test_stream, model=model)
 141
 142         self.failUnlessRaises(daf.ModelException,
 143                               daf_mapper.find_view,
 144                               'filename_r1.fastq')
 145
 146     def test_find_attributes(self):
 147         lib_id = '11204'
 148         lib_url = 'http://jumpgate.caltech.edu/library/%s' %(lib_id)
 149         model = get_model()
 150
 151         parser = RDF.Parser(name='turtle')
 152         parser.parse_string_into_model(model, '''
 153 @prefix dafTerm: <http://jumpgate.caltech.edu/wiki/UcscDaf#> .
 154 @prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
 155
 156 <%(submissionLog)s/testfind/view/Signal> dafTerm:filename_re ".*\\\\.bam" .
 157 <%(submissionLog)s/testfind/view/FastqRd1> dafTerm:filename_re ".*\\\\.fastq" .
 158 <%(libUrl)s> <%(libraryOntology)sgel_cut> "100"^^xsd:decimal .
 159 ''' % {'submissionLog': 'http://jumpgate.caltech.edu/wiki/SubmissionsLog',
 160        'libraryOntology': 'http://jumpgate.caltech.edu/wiki/LibraryOntology#',
 161        'libUrl': lib_url},
 162        'http://blank')
 163         name = 'testfind'
 164         test_stream = StringIO(test_daf)
 165         daf_mapper = daf.DAFMapper(name, daf_file=test_stream, model=model)
 166         libNode = RDF.Node(RDF.Uri(lib_url))
 167         daf_mapper._add_library_details_to_model(libNode)
 168         gel_cut = daf_mapper._get_library_attribute(libNode, 'gel_cut')
 169         # make sure we can override attributes, the value in our
 170         # server is 500 for this library
 171         self.failUnlessEqual(gel_cut, 100)
 172
 173         species = daf_mapper._get_library_attribute(libNode, 'species')
 174         self.failUnlessEqual(species, "Homo sapiens")
 175
 176         daf_mapper.construct_file_attributes('/tmp/analysis1', libNode, 'filename.bam')
 177         source = daf_mapper.model.get_source(rdfNS['type'], submissionOntology['submission'])
 178         self.failUnlessEqual(str(source), "<http://jumpgate.caltech.edu/wiki/SubmissionsLog/testfind/analysis1>")
 179         view = daf_mapper.model.get_target(source, submissionOntology['has_view'])
 180         self.failUnlessEqual(str(view), "<http://jumpgate.caltech.edu/wiki/SubmissionsLog/testfind/view/Signal>")
 181
 182 def suite():
 183     suite = unittest.makeSuite(TestDAF, 'test')
 184     suite.addTest(unittest.makeSuite(TestDAFMapper, 'test'))
 185     return suite
 186
 187 if __name__ == "__main__":
 188     unittest.main(defaultTest='suite')