htsworkflow/pipelines/test/test_runfolder110.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 firecrest
  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                                  '081017_HWI-EAS229_0062_30J55AAXX')
  27     os.mkdir(runfolder_dir)
  28
  29     data_dir = os.path.join(runfolder_dir, 'Data')
  30     os.mkdir(data_dir)
  31
  32     firecrest_dir = os.path.join(data_dir,
  33                                  'C1-37_Firecrest1.9.6_20-10-2008_diane')
  34     os.mkdir(firecrest_dir)
  35
  36     bustard_dir = os.path.join(firecrest_dir,
  37                                'Bustard1.9.6_20-10-2008_diane')
  38     os.mkdir(bustard_dir)
  39     make_phasing_params(bustard_dir)
  40
  41     matrix_name = os.path.join(bustard_dir, 'matrix1.txt')
  42     make_matrix(matrix_name)
  43
  44
  45     gerald_dir = os.path.join(bustard_dir,
  46                               'GERALD_20-10-2008_diane')
  47     os.mkdir(gerald_dir)
  48     make_gerald_config_100(gerald_dir)
  49     make_summary_htm_110(gerald_dir)
  50     make_eland_multi(gerald_dir)
  51
  52     if obj is not None:
  53         obj.temp_dir = temp_dir
  54         obj.runfolder_dir = runfolder_dir
  55         obj.data_dir = data_dir
  56         obj.image_analysis_dir = firecrest_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_firecrest(self):
  74         """
  75         Construct a firecrest object
  76         """
  77         f = firecrest.firecrest(self.image_analysis_dir)
  78         self.failUnlessEqual(f.software, 'Firecrest')
  79         self.failUnlessEqual(f.version, '1.9.6')
  80         self.failUnlessEqual(f.start, 1)
  81         self.failUnlessEqual(f.stop, 37)
  82         self.failUnlessEqual(f.user, 'diane')
  83         self.failUnlessEqual(f.date, date(2008,10,20))
  84
  85         xml = f.get_elements()
  86         # just make sure that element tree can serialize the tree
  87         xml_str = ElementTree.tostring(xml)
  88
  89         f2 = firecrest.Firecrest(xml=xml)
  90         self.failUnlessEqual(f.software, f2.software)
  91         self.failUnlessEqual(f.version, f2.version)
  92         self.failUnlessEqual(f.start,   f2.start)
  93         self.failUnlessEqual(f.stop,    f2.stop)
  94         self.failUnlessEqual(f.user,    f2.user)
  95
  96     def test_bustard(self):
  97         """
  98         construct a bustard object
  99         """
 100         b = bustard.bustard(self.bustard_dir)
 101         self.failUnlessEqual(b.software, 'Bustard')
 102         self.failUnlessEqual(b.version, '1.9.6')
 103         self.failUnlessEqual(b.date,    date(2008,10,20))
 104         self.failUnlessEqual(b.user,    'diane')
 105         self.failUnlessEqual(len(b.phasing), 8)
 106         self.failUnlessAlmostEqual(b.phasing[8].phasing, 0.0099)
 107
 108         xml = b.get_elements()
 109         b2 = bustard.Bustard(xml=xml)
 110         self.failUnlessEqual(b.software, b2.software)
 111         self.failUnlessEqual(b.version, b2.version)
 112         self.failUnlessEqual(b.date,    b2.date )
 113         self.failUnlessEqual(b.user,    b2.user)
 114         self.failUnlessEqual(len(b.phasing), len(b2.phasing))
 115         for key in b.phasing.keys():
 116             self.failUnlessEqual(b.phasing[key].lane,
 117                                  b2.phasing[key].lane)
 118             self.failUnlessEqual(b.phasing[key].phasing,
 119                                  b2.phasing[key].phasing)
 120             self.failUnlessEqual(b.phasing[key].prephasing,
 121                                  b2.phasing[key].prephasing)
 122
 123     def test_gerald(self):
 124         # need to update gerald and make tests for it
 125         g = gerald.gerald(self.gerald_dir)
 126
 127         self.failUnlessEqual(g.software, 'GERALD')
 128         self.failUnlessEqual(g.version, '1.171')
 129         self.failUnlessEqual(g.date, datetime(2009,2,22,21,15,59))
 130         self.failUnlessEqual(len(g.lanes), len(g.lanes.keys()))
 131         self.failUnlessEqual(len(g.lanes), len(g.lanes.items()))
 132
 133
 134         # list of genomes, matches what was defined up in
 135         # make_gerald_config.
 136         # the first None is to offset the genomes list to be 1..9
 137         # instead of pythons default 0..8
 138         genomes = [None,
 139                    '/g/mm9',
 140                    '/g/mm9',
 141                    '/g/elegans190',
 142                    '/g/arabidopsis01222004',
 143                    '/g/mm9',
 144                    '/g/mm9',
 145                    '/g/mm9',
 146                    '/g/mm9', ]
 147
 148         # test lane specific parameters from gerald config file
 149         for i in range(1,9):
 150             cur_lane = g.lanes[i]
 151             self.failUnlessEqual(cur_lane.analysis, 'eland_extended')
 152             self.failUnlessEqual(cur_lane.eland_genome, genomes[i])
 153             self.failUnlessEqual(cur_lane.read_length, '37')
 154             self.failUnlessEqual(cur_lane.use_bases, 'Y'*37)
 155
 156         # I want to be able to use a simple iterator
 157         for l in g.lanes.values():
 158           self.failUnlessEqual(l.analysis, 'eland_extended')
 159           self.failUnlessEqual(l.read_length, '37')
 160           self.failUnlessEqual(l.use_bases, 'Y'*37)
 161
 162         # raw cluster numbers extracted from summary file
 163         # its the first +/- value in the lane results summary
 164         # section
 165         clusters = [None,
 166                     (190220, 15118), (190560, 14399),
 167                     (187597, 12369), (204142, 16877),
 168                     (247308, 11600), (204298, 15640),
 169                     (202707, 15404), (198075, 14702),]
 170
 171         self.failUnlessEqual(len(g.summary), 1)
 172         for i in range(1,9):
 173             summary_lane = g.summary[0][i]
 174             self.failUnlessEqual(summary_lane.cluster, clusters[i])
 175             self.failUnlessEqual(summary_lane.lane, i)
 176
 177         xml = g.get_elements()
 178         # just make sure that element tree can serialize the tree
 179         xml_str = ElementTree.tostring(xml)
 180         g2 = gerald.Gerald(xml=xml)
 181
 182         # do it all again after extracting from the xml file
 183         self.failUnlessEqual(g.software, g2.software)
 184         self.failUnlessEqual(g.version, g2.version)
 185         self.failUnlessEqual(g.date, g2.date)
 186         self.failUnlessEqual(len(g.lanes.keys()), len(g2.lanes.keys()))
 187         self.failUnlessEqual(len(g.lanes.items()), len(g2.lanes.items()))
 188
 189         # test lane specific parameters from gerald config file
 190         for i in range(1,9):
 191             g_lane = g.lanes[i]
 192             g2_lane = g2.lanes[i]
 193             self.failUnlessEqual(g_lane.analysis, g2_lane.analysis)
 194             self.failUnlessEqual(g_lane.eland_genome, g2_lane.eland_genome)
 195             self.failUnlessEqual(g_lane.read_length, g2_lane.read_length)
 196             self.failUnlessEqual(g_lane.use_bases, g2_lane.use_bases)
 197
 198         self.failUnlessEqual(len(g.summary), 1)
 199         # test (some) summary elements
 200         for i in range(1,9):
 201             g_summary = g.summary[0][i]
 202             g2_summary = g2.summary[0][i]
 203             self.failUnlessEqual(g_summary.cluster, g2_summary.cluster)
 204             self.failUnlessEqual(g_summary.lane, g2_summary.lane)
 205
 206             g_eland = g.eland_results
 207             g2_eland = g2.eland_results
 208             for key in g_eland:
 209                 g_results = g_eland[key]
 210                 g2_results = g2_eland[key]
 211                 self.failUnlessEqual(g_results.reads,
 212                                      g2_results.reads)
 213                 self.failUnlessEqual(len(g_results.mapped_reads),
 214                                      len(g2_results.mapped_reads))
 215                 for k in g_results.mapped_reads.keys():
 216                     self.failUnlessEqual(g_results.mapped_reads[k],
 217                                          g2_results.mapped_reads[k])
 218
 219                 self.failUnlessEqual(len(g_results.match_codes),
 220                                      len(g2_results.match_codes))
 221                 for k in g_results.match_codes.keys():
 222                     self.failUnlessEqual(g_results.match_codes[k],
 223                                          g2_results.match_codes[k])
 224
 225
 226     def test_eland(self):
 227         hg_map = {'Lambda.fa': 'Lambda.fa'}
 228         for i in range(1,22):
 229           short_name = 'chr%d.fa' % (i,)
 230           long_name = 'hg18/chr%d.fa' % (i,)
 231           hg_map[short_name] = long_name
 232
 233         genome_maps = { 1:hg_map, 2:hg_map, 3:hg_map, 4:hg_map,
 234                         5:hg_map, 6:hg_map, 7:hg_map, 8:hg_map }
 235         eland = gerald.eland(self.gerald_dir, genome_maps=genome_maps)
 236
 237         for key in eland:
 238             lane = eland[key]
 239             self.failUnlessEqual(lane.reads, 6)
 240             self.failUnlessEqual(lane.sample_name, "s")
 241             self.failUnlessEqual(lane.lane_id, key.lane)
 242             self.failUnlessEqual(len(lane.mapped_reads), 17)
 243             self.failUnlessEqual(lane.mapped_reads['hg18/chr5.fa'], 4)
 244             self.failUnlessEqual(lane.match_codes['U0'], 3)
 245             self.failUnlessEqual(lane.match_codes['R0'], 2)
 246             self.failUnlessEqual(lane.match_codes['U1'], 1)
 247             self.failUnlessEqual(lane.match_codes['R1'], 9)
 248             self.failUnlessEqual(lane.match_codes['U2'], 0)
 249             self.failUnlessEqual(lane.match_codes['R2'], 12)
 250             self.failUnlessEqual(lane.match_codes['NM'], 1)
 251             self.failUnlessEqual(lane.match_codes['QC'], 0)
 252
 253         xml = eland.get_elements()
 254         # just make sure that element tree can serialize the tree
 255         xml_str = ElementTree.tostring(xml)
 256         e2 = gerald.ELAND(xml=xml)
 257
 258         for key in eland:
 259             l1 = eland[key]
 260             l2 = e2[key]
 261             self.failUnlessEqual(l1.reads, l2.reads)
 262             self.failUnlessEqual(l1.sample_name, l2.sample_name)
 263             self.failUnlessEqual(l1.lane_id, l2.lane_id)
 264             self.failUnlessEqual(len(l1.mapped_reads), len(l2.mapped_reads))
 265             self.failUnlessEqual(len(l1.mapped_reads), 17)
 266             for k in l1.mapped_reads.keys():
 267                 self.failUnlessEqual(l1.mapped_reads[k],
 268                                      l2.mapped_reads[k])
 269
 270             self.failUnlessEqual(len(l1.match_codes), 9)
 271             self.failUnlessEqual(len(l1.match_codes), len(l2.match_codes))
 272             for k in l1.match_codes.keys():
 273                 self.failUnlessEqual(l1.match_codes[k],
 274                                      l2.match_codes[k])
 275
 276     def test_runfolder(self):
 277         runs = runfolder.get_runs(self.runfolder_dir)
 278
 279         # do we get the flowcell id from the filename?
 280         self.failUnlessEqual(len(runs), 1)
 281         name = 'run_30J55AAXX_2009-02-22.xml'
 282         self.failUnlessEqual(runs[0].serialization_filename, name)
 283
 284         # do we get the flowcell id from the FlowcellId.xml file
 285         make_flowcell_id(self.runfolder_dir, '30J55AAXX')
 286         runs = runfolder.get_runs(self.runfolder_dir)
 287         self.failUnlessEqual(len(runs), 1)
 288         name = 'run_30J55AAXX_2009-02-22.xml'
 289         self.failUnlessEqual(runs[0].serialization_filename, name)
 290
 291         r1 = runs[0]
 292         xml = r1.get_elements()
 293         xml_str = ElementTree.tostring(xml)
 294
 295         r2 = runfolder.PipelineRun(xml=xml)
 296         self.failUnlessEqual(r1.serialization_filename, r2.serialization_filename)
 297         self.failIfEqual(r2.image_analysis, None)
 298         self.failIfEqual(r2.bustard, None)
 299         self.failIfEqual(r2.gerald, None)
 300
 301
 302 def suite():
 303     from unittest import TestSuite, defaultTestLoader
 304     suite = TestSuite()
 305     suite.addTests(defaultTestLoader.loadTestsFromTestCase(RunfolderTests))
 306     return suite
 307
 308
 309 if __name__ == "__main__":
 310     from unittest import main
 311     main(defaultTest="suite")