htsworkflow/pipelines/test/test_runfolder_ipar100.py

   1 #!/usr/bin/env python
   2
   3 from datetime import datetime, date
   4 import os
   5 import tempfile
   6 import shutil
   7 from unittest import TestCase
   8
   9 from htsworkflow.pipelines import ipar
  10 from htsworkflow.pipelines import bustard
  11 from htsworkflow.pipelines import gerald
  12 from htsworkflow.pipelines import runfolder
  13 from htsworkflow.pipelines import ElementTree
  14
  15 from htsworkflow.pipelines.test.simulate_runfolder import *
  16
  17
  18 def make_runfolder(obj=None):
  19     """
  20     Make a fake runfolder, attach all the directories to obj if defined
  21     """
  22     # make a fake runfolder directory
  23     temp_dir = tempfile.mkdtemp(prefix='tmp_runfolder_')
  24
  25     runfolder_dir = os.path.join(temp_dir,
  26                                  '080102_HWI-EAS229_0010_207BTAAXX')
  27     os.mkdir(runfolder_dir)
  28
  29     data_dir = os.path.join(runfolder_dir, 'Data')
  30     os.mkdir(data_dir)
  31
  32     ipar_dir = make_ipar_dir(data_dir)
  33
  34     matrix_dir = os.path.join(ipar_dir, 'Matrix')
  35     os.mkdir(matrix_dir)
  36     matrix_name = os.path.join(matrix_dir, 's_matrix.txt')
  37     make_matrix(matrix_name)
  38
  39     bustard_dir = os.path.join(ipar_dir,
  40                                'Bustard1.8.28_12-04-2008_diane')
  41     os.mkdir(bustard_dir)
  42     make_phasing_params(bustard_dir)
  43
  44     gerald_dir = os.path.join(bustard_dir,
  45                               'GERALD_12-04-2008_diane')
  46     os.mkdir(gerald_dir)
  47     make_gerald_config_100(gerald_dir)
  48     make_summary_htm_100(gerald_dir)
  49     make_eland_multi(gerald_dir)
  50
  51     if obj is not None:
  52         obj.temp_dir = temp_dir
  53         obj.runfolder_dir = runfolder_dir
  54         obj.data_dir = data_dir
  55         obj.image_analysis_dir = ipar_dir
  56         obj.matrix_dir = matrix_dir
  57         obj.bustard_dir = bustard_dir
  58         obj.gerald_dir = gerald_dir
  59
  60
  61 class RunfolderTests(TestCase):
  62     """
  63     Test components of the runfolder processing code
  64     which includes firecrest, bustard, and gerald
  65     """
  66     def setUp(self):
  67         # attaches all the directories to the object passed in
  68         make_runfolder(self)
  69
  70     def tearDown(self):
  71         shutil.rmtree(self.temp_dir)
  72
  73     def test_ipar(self):
  74         """
  75         Construct a firecrest object
  76         """
  77         i = ipar.ipar(self.image_analysis_dir)
  78         self.failUnlessEqual(i.software, 'IPAR')
  79         self.failUnlessEqual(i.version, '2.01.192.0')
  80         self.failUnlessEqual(i.start, 1)
  81         self.failUnlessEqual(i.stop, 37)
  82
  83         xml = i.get_elements()
  84         # just make sure that element tree can serialize the tree
  85         xml_str = ElementTree.tostring(xml)
  86
  87         i2 = ipar.IPAR(xml=xml)
  88         self.failUnlessEqual(i.software, i2.software)
  89         self.failUnlessEqual(i.version, i2.version)
  90         self.failUnlessEqual(i.start,   i2.start)
  91         self.failUnlessEqual(i.stop,    i2.stop)
  92         self.failUnlessEqual(i.date,    i2.date)
  93         self.failUnlessEqual(i.file_list(), i2.file_list())
  94
  95     def test_bustard(self):
  96         """
  97         construct a bustard object
  98         """
  99         b = bustard.bustard(self.bustard_dir)
 100         self.failUnlessEqual(b.software, 'Bustard')
 101         self.failUnlessEqual(b.version, '1.8.28')
 102         self.failUnlessEqual(b.date,    date(2008,4,12))
 103         self.failUnlessEqual(b.user,    'diane')
 104         self.failUnlessEqual(len(b.phasing), 8)
 105         self.failUnlessAlmostEqual(b.phasing[8].phasing, 0.0099)
 106
 107         xml = b.get_elements()
 108         b2 = bustard.Bustard(xml=xml)
 109         self.failUnlessEqual(b.software, b2.software)
 110         self.failUnlessEqual(b.version, b2.version)
 111         self.failUnlessEqual(b.date,    b2.date )
 112         self.failUnlessEqual(b.user,    b2.user)
 113         self.failUnlessEqual(len(b.phasing), len(b2.phasing))
 114         for key in b.phasing.keys():
 115             self.failUnlessEqual(b.phasing[key].lane,
 116                                  b2.phasing[key].lane)
 117             self.failUnlessEqual(b.phasing[key].phasing,
 118                                  b2.phasing[key].phasing)
 119             self.failUnlessEqual(b.phasing[key].prephasing,
 120                                  b2.phasing[key].prephasing)
 121
 122     def test_gerald(self):
 123         # need to update gerald and make tests for it
 124         g = gerald.gerald(self.gerald_dir)
 125
 126         self.failUnlessEqual(g.software, 'GERALD')
 127         self.failUnlessEqual(g.version, '1.171')
 128         self.failUnlessEqual(g.date, datetime(2009,2,22,21,15,59))
 129         self.failUnlessEqual(len(g.lanes), len(g.lanes.keys()))
 130         self.failUnlessEqual(len(g.lanes), len(g.lanes.items()))
 131
 132
 133         # list of genomes, matches what was defined up in
 134         # make_gerald_config.
 135         # the first None is to offset the genomes list to be 1..9
 136         # instead of pythons default 0..8
 137         genomes = [None,
 138                    '/g/mm9',
 139                    '/g/mm9',
 140                    '/g/elegans190',
 141                    '/g/arabidopsis01222004',
 142                    '/g/mm9',
 143                    '/g/mm9',
 144                    '/g/mm9',
 145                    '/g/mm9', ]
 146
 147         # test lane specific parameters from gerald config file
 148         for i in range(1,9):
 149             cur_lane = g.lanes[i]
 150             self.failUnlessEqual(cur_lane.analysis, 'eland_extended')
 151             self.failUnlessEqual(cur_lane.eland_genome, genomes[i])
 152             self.failUnlessEqual(cur_lane.read_length, '37')
 153             self.failUnlessEqual(cur_lane.use_bases, 'Y'*37)
 154
 155         # I want to be able to use a simple iterator
 156         for l in g.lanes.values():
 157           self.failUnlessEqual(l.analysis, 'eland_extended')
 158           self.failUnlessEqual(l.read_length, '37')
 159           self.failUnlessEqual(l.use_bases, 'Y'*37)
 160
 161         # test data extracted from summary file
 162         clusters = [None,
 163                     (96483, 9074), (133738, 7938),
 164                     (152142, 10002), (15784, 2162),
 165                     (119735, 8465), (152177, 8146),
 166                     (84649, 7325), (54622, 4812),]
 167
 168         self.failUnlessEqual(len(g.summary), 1)
 169         for i in range(1,9):
 170             summary_lane = g.summary[0][i]
 171             self.failUnlessEqual(summary_lane.cluster, clusters[i])
 172             self.failUnlessEqual(summary_lane.lane, i)
 173
 174         xml = g.get_elements()
 175         # just make sure that element tree can serialize the tree
 176         xml_str = ElementTree.tostring(xml)
 177         g2 = gerald.Gerald(xml=xml)
 178
 179         # do it all again after extracting from the xml file
 180         self.failUnlessEqual(g.software, g2.software)
 181         self.failUnlessEqual(g.version, g2.version)
 182         self.failUnlessEqual(g.date, g2.date)
 183         self.failUnlessEqual(len(g.lanes.keys()), len(g2.lanes.keys()))
 184         self.failUnlessEqual(len(g.lanes.items()), len(g2.lanes.items()))
 185
 186         # test lane specific parameters from gerald config file
 187         for i in range(1,9):
 188             g_lane = g.lanes[i]
 189             g2_lane = g2.lanes[i]
 190             self.failUnlessEqual(g_lane.analysis, g2_lane.analysis)
 191             self.failUnlessEqual(g_lane.eland_genome, g2_lane.eland_genome)
 192             self.failUnlessEqual(g_lane.read_length, g2_lane.read_length)
 193             self.failUnlessEqual(g_lane.use_bases, g2_lane.use_bases)
 194
 195         # test (some) summary elements
 196         self.failUnlessEqual(len(g.summary), 1)
 197         for i in range(1,9):
 198             g_summary = g.summary[0][i]
 199             g2_summary = g2.summary[0][i]
 200             self.failUnlessEqual(g_summary.cluster, g2_summary.cluster)
 201             self.failUnlessEqual(g_summary.lane, g2_summary.lane)
 202
 203             g_eland = g.eland_results
 204             g2_eland = g2.eland_results
 205             for key in g_eland:
 206                 g_results = g_eland[key]
 207                 g2_results = g2_eland[key]
 208                 self.failUnlessEqual(g_results.reads,
 209                                      g2_results.reads)
 210                 self.failUnlessEqual(len(g_results.mapped_reads),
 211                                      len(g2_results.mapped_reads))
 212                 for k in g_results.mapped_reads.keys():
 213                     self.failUnlessEqual(g_results.mapped_reads[k],
 214                                          g2_results.mapped_reads[k])
 215
 216                 self.failUnlessEqual(len(g_results.match_codes),
 217                                      len(g2_results.match_codes))
 218                 for k in g_results.match_codes.keys():
 219                     self.failUnlessEqual(g_results.match_codes[k],
 220                                          g2_results.match_codes[k])
 221
 222
 223     def test_eland(self):
 224         hg_map = {'Lambda.fa': 'Lambda.fa'}
 225         for i in range(1,22):
 226           short_name = 'chr%d.fa' % (i,)
 227           long_name = 'hg18/chr%d.fa' % (i,)
 228           hg_map[short_name] = long_name
 229
 230         genome_maps = { 1:hg_map, 2:hg_map, 3:hg_map, 4:hg_map,
 231                         5:hg_map, 6:hg_map, 7:hg_map, 8:hg_map }
 232         eland = gerald.eland(self.gerald_dir, genome_maps=genome_maps)
 233
 234         for key in eland:
 235             lane = eland[key]
 236             self.failUnlessEqual(lane.reads, 6)
 237             self.failUnlessEqual(lane.sample_name, "s")
 238             self.failUnlessEqual(lane.lane_id, key.lane)
 239             self.failUnlessEqual(len(lane.mapped_reads), 17)
 240             self.failUnlessEqual(lane.mapped_reads['hg18/chr5.fa'], 4)
 241             self.failUnlessEqual(lane.mapped_reads['spike.fa/sample1'], 1)
 242             self.failUnlessEqual(lane.mapped_reads['spike.fa/sample2'], 1)
 243             self.failUnlessEqual(lane.match_codes['U0'], 3)
 244             self.failUnlessEqual(lane.match_codes['R0'], 2)
 245             self.failUnlessEqual(lane.match_codes['U1'], 1)
 246             self.failUnlessEqual(lane.match_codes['R1'], 9)
 247             self.failUnlessEqual(lane.match_codes['U2'], 0)
 248             self.failUnlessEqual(lane.match_codes['R2'], 12)
 249             self.failUnlessEqual(lane.match_codes['NM'], 1)
 250             self.failUnlessEqual(lane.match_codes['QC'], 0)
 251
 252         xml = eland.get_elements()
 253         # just make sure that element tree can serialize the tree
 254         xml_str = ElementTree.tostring(xml)
 255         e2 = gerald.ELAND(xml=xml)
 256
 257         for key in eland:
 258             l1 = eland[key]
 259             l2 = e2[key]
 260             self.failUnlessEqual(l1.reads, l2.reads)
 261             self.failUnlessEqual(l1.sample_name, l2.sample_name)
 262             self.failUnlessEqual(l1.lane_id, l2.lane_id)
 263             self.failUnlessEqual(len(l1.mapped_reads), len(l2.mapped_reads))
 264             self.failUnlessEqual(len(l1.mapped_reads), 17)
 265             for k in l1.mapped_reads.keys():
 266                 self.failUnlessEqual(l1.mapped_reads[k],
 267                                      l2.mapped_reads[k])
 268
 269             self.failUnlessEqual(len(l1.match_codes), 9)
 270             self.failUnlessEqual(len(l1.match_codes), len(l2.match_codes))
 271             for k in l1.match_codes.keys():
 272                 self.failUnlessEqual(l1.match_codes[k],
 273                                      l2.match_codes[k])
 274
 275     def test_runfolder(self):
 276         runs = runfolder.get_runs(self.runfolder_dir)
 277
 278         # do we get the flowcell id from the filename?
 279         self.failUnlessEqual(len(runs), 1)
 280         name = 'run_207BTAAXX_%s.xml' % ( date.today().strftime('%Y-%m-%d'),)
 281         self.failUnlessEqual(runs[0].serialization_filename, name)
 282
 283         # do we get the flowcell id from the FlowcellId.xml file
 284         make_flowcell_id(self.runfolder_dir, '207BTAAXY')
 285         runs = runfolder.get_runs(self.runfolder_dir)
 286         self.failUnlessEqual(len(runs), 1)
 287         name = 'run_207BTAAXY_%s.xml' % ( date.today().strftime('%Y-%m-%d'),)
 288         self.failUnlessEqual(runs[0].serialization_filename, name)
 289
 290         r1 = runs[0]
 291         xml = r1.get_elements()
 292         xml_str = ElementTree.tostring(xml)
 293
 294         r2 = runfolder.PipelineRun(xml=xml)
 295         self.failUnlessEqual(r1.serialization_filename, r2.serialization_filename)
 296         self.failIfEqual(r2.image_analysis, None)
 297         self.failIfEqual(r2.bustard, None)
 298         self.failIfEqual(r2.gerald, None)
 299
 300
 301 def suite():
 302     from unittest import TestSuite, defaultTestLoader
 303     suite = TestSuite()
 304     suite.addTests(defaultTestLoader.loadTestsFromTestCase(RunfolderTests))
 305     return suite
 306
 307
 308 if __name__ == "__main__":
 309     from unittest import main
 310     main(defaultTest="suite")