htsworkflow/pipelines/test/test_genomemap.py

   1 #!/usr/bin/env python
   2 """More direct synthetic test cases for the eland output file processing
   3 """
   4 import os
   5 from StringIO import StringIO
   6 import shutil
   7 import tempfile
   8 from unittest import TestCase
   9
  10 from htsworkflow.pipelines import ElementTree
  11 from htsworkflow.pipelines import genomemap
  12
  13 MINI_GENOME_XML = '''<sequenceSizes>
  14         <chromosome fileName="chr2.fa" contigName="chr2" totalBases="181748087"/>
  15         <chromosome fileName="chr1.fa" contigName="chr1" totalBases="197195432"/>
  16 </sequenceSizes>
  17 '''
  18 class TestGenomeMap(TestCase):
  19     def test_genomesizes_xml(self):
  20         xml = ElementTree.fromstring(MINI_GENOME_XML)
  21         g = genomemap.GenomeMap()
  22         g.build_map_from_element(xml)
  23
  24         self.assertTrue('chr1.fa' in g)
  25         self.assertEqual(g['chr1.fa'], 'mm9/chr1.fa')
  26
  27     def test_genomesizes_file(self):
  28         g = genomemap.GenomeMap()
  29         try:
  30             tempdir = tempfile.mkdtemp(prefix='tmp_genome')
  31             name = os.path.join(tempdir, '11111_NoIndex_L001_genomesize.xml')
  32             stream = open(name, 'w')
  33             stream.write(MINI_GENOME_XML)
  34             stream.close()
  35             g.parse_genomesize(name)
  36         finally:
  37             shutil.rmtree(tempdir)
  38
  39         self.assertTrue('chr1.fa' in g)
  40         self.assertEqual(len(g), 2)
  41         self.assertEqual(g['chr1.fa'], 'mm9/chr1.fa')
  42
  43     def test_simulated_genome_dir(self):
  44         vlds = [genomemap.vldInfo('chr1.fa.vld', False),
  45                 genomemap.vldInfo('chr2.fa.vld', False),
  46                 genomemap.vldInfo('chr3.fa.vld', False),
  47                 genomemap.vldInfo('Lambda.fa.vld', True),]
  48
  49         g = genomemap.GenomeMap()
  50         g.build_map_from_dir('mm9', vlds)
  51
  52         self.assertTrue('chr1.fa' in g)
  53         self.assertEqual(len(g), 4)
  54         self.assertEqual(g['chr1.fa'], 'mm9/chr1.fa')
  55         self.assertEqual(g['Lambda.fa'], 'Lambda.fa')
  56
  57     def test_genome_dir(self):
  58         g = genomemap.GenomeMap()
  59         names = ['chr1', 'chr2', 'chr3']
  60         tempdir = None
  61         try:
  62             tempdir = tempfile.mkdtemp(prefix='tmp_mm9')
  63             for base in names:
  64                 name = os.path.join(tempdir, base + '.fa.vld')
  65                 stream = open(name, 'w')
  66                 stream.write(name)
  67                 stream.close()
  68             g.scan_genome_dir(tempdir)
  69         finally:
  70             if tempdir is not None:
  71                 shutil.rmtree(tempdir)
  72
  73         temppath, tempgenome = os.path.split(tempdir)
  74         self.assertTrue('chr1.fa' in g)
  75         self.assertEqual(len(g), 3)
  76         self.assertEqual(g['chr1.fa'], '{0}/chr1.fa'.format(tempgenome))
  77
  78
  79 def suite():
  80     from unittest import TestSuite, defaultTestLoader
  81     suite = TestSuite()
  82     suite.addTests(defaultTestLoader.loadTestsFromTestCase(TestGenomeMap))
  83     return suite
  84
  85
  86 if __name__ == "__main__":
  87     from unittest import main
  88     main(defaultTest="suite")