htsworkflow/pipelines/test/test_runfolder_pair.py

   1 #!/usr/bin/env python
   2
   3 from datetime import datetime, date
   4 import os
   5 import tempfile
   6 import shutil
   7 import unittest
   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.runfolder 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_firecrest_dir(data_dir, "1.9.6", 1, 152)
  33
  34     matrix_dir = os.path.join(ipar_dir, 'Matrix')
  35     os.mkdir(matrix_dir)
  36     make_matrix(matrix_dir)
  37
  38     bustard_dir = os.path.join(ipar_dir,
  39                                'Bustard1.8.28_12-04-2008_diane')
  40     os.mkdir(bustard_dir)
  41     make_phasing_params(bustard_dir)
  42
  43     gerald_dir = os.path.join(bustard_dir,
  44                               'GERALD_12-04-2008_diane')
  45     os.mkdir(gerald_dir)
  46     make_gerald_config(gerald_dir)
  47     make_summary_paired_htm(gerald_dir)
  48     make_eland_multi(gerald_dir)
  49
  50     if obj is not None:
  51         obj.temp_dir = temp_dir
  52         obj.runfolder_dir = runfolder_dir
  53         obj.data_dir = data_dir
  54         obj.image_analysis_dir = ipar_dir
  55         obj.matrix_dir = matrix_dir
  56         obj.bustard_dir = bustard_dir
  57         obj.gerald_dir = gerald_dir
  58
  59
  60 class RunfolderTests(unittest.TestCase):
  61     """
  62     Test components of the runfolder processing code
  63     which includes firecrest, bustard, and gerald
  64     """
  65     def setUp(self):
  66         # attaches all the directories to the object passed in
  67         make_runfolder(self)
  68
  69     def tearDown(self):
  70         shutil.rmtree(self.temp_dir)
  71
  72     def test_firecrest(self):
  73         """
  74         Construct a firecrest object
  75         """
  76         f = firecrest.firecrest(self.image_analysis_dir)
  77         self.failUnlessEqual(f.version, '1.9.6')
  78         self.failUnlessEqual(f.start, 1)
  79         self.failUnlessEqual(f.stop, 152)
  80         self.failUnlessEqual(f.user, 'diane')
  81         # As of 2008-12-8, the date was being set in
  82         # simulate_runfolder.make_firecrest_dir
  83         self.failUnlessEqual(f.date, date(2008,4,12))
  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.version, f2.version)
  91         self.failUnlessEqual(f.start,   f2.start)
  92         self.failUnlessEqual(f.stop,    f2.stop)
  93         self.failUnlessEqual(f.user,    f2.user)
  94
  95     def test_bustard(self):
  96         """
  97         construct a bustard object
  98         """
  99         b = bustard.bustard(self.bustard_dir)
 100         self.failUnlessEqual(b.version, '1.8.28')
 101         self.failUnlessEqual(b.date,    date(2008,4,12))
 102         self.failUnlessEqual(b.user,    'diane')
 103         self.failUnlessEqual(len(b.phasing), 8)
 104         self.failUnlessAlmostEqual(b.phasing[8].phasing, 0.0099)
 105
 106         xml = b.get_elements()
 107         b2 = bustard.Bustard(xml=xml)
 108         self.failUnlessEqual(b.version, b2.version)
 109         self.failUnlessEqual(b.date,    b2.date )
 110         self.failUnlessEqual(b.user,    b2.user)
 111         self.failUnlessEqual(len(b.phasing), len(b2.phasing))
 112         for key in b.phasing.keys():
 113             self.failUnlessEqual(b.phasing[key].lane,
 114                                  b2.phasing[key].lane)
 115             self.failUnlessEqual(b.phasing[key].phasing,
 116                                  b2.phasing[key].phasing)
 117             self.failUnlessEqual(b.phasing[key].prephasing,
 118                                  b2.phasing[key].prephasing)
 119
 120     def test_gerald(self):
 121         # need to update gerald and make tests for it
 122         g = gerald.gerald(self.gerald_dir)
 123
 124         self.failUnlessEqual(g.version,
 125             '@(#) Id: GERALD.pl,v 1.68.2.2 2007/06/13 11:08:49 km Exp')
 126         self.failUnlessEqual(g.date, datetime(2008,4,19,19,8,30))
 127         self.failUnlessEqual(len(g.lanes), len(g.lanes.keys()))
 128         self.failUnlessEqual(len(g.lanes), len(g.lanes.items()))
 129
 130
 131         # list of genomes, matches what was defined up in
 132         # make_gerald_config.
 133         # the first None is to offset the genomes list to be 1..9
 134         # instead of pythons default 0..8
 135         genomes = [None, '/g/dm3', '/g/equcab1', '/g/equcab1', '/g/canfam2',
 136                          '/g/hg18', '/g/hg18', '/g/hg18', '/g/hg18', ]
 137
 138         # test lane specific parameters from gerald config file
 139         for i in range(1,9):
 140             cur_lane = g.lanes[i]
 141             self.failUnlessEqual(cur_lane.analysis, 'eland')
 142             self.failUnlessEqual(cur_lane.eland_genome, genomes[i])
 143             self.failUnlessEqual(cur_lane.read_length, '32')
 144             self.failUnlessEqual(cur_lane.use_bases, 'Y'*32)
 145
 146         # I want to be able to use a simple iterator
 147         for l in g.lanes.values():
 148           self.failUnlessEqual(l.analysis, 'eland')
 149           self.failUnlessEqual(l.read_length, '32')
 150           self.failUnlessEqual(l.use_bases, 'Y'*32)
 151
 152         # test data extracted from summary file
 153         clusters = [[None,
 154                     (103646, 4515), (106678, 4652),
 155                     (84583, 5963), (68813, 4782),
 156                     (104854, 4664), (43555, 1632),
 157                     (54265, 1588), (64363, 2697),],
 158                     [None,
 159                     (103647, 4516), (106679, 4653),
 160                     (84584, 5964), (68814, 4783),
 161                     (104855, 4665), (43556, 1633),
 162                     (54266, 1589), (64364, 2698),],]
 163
 164         for end in [0,1]:
 165             for lane in range(1,9):
 166                 summary_lane = g.summary[end][lane]
 167                 self.failUnlessEqual(summary_lane.cluster, clusters[end][lane])
 168                 self.failUnlessEqual(summary_lane.lane, lane)
 169
 170         xml = g.get_elements()
 171         # just make sure that element tree can serialize the tree
 172         xml_str = ElementTree.tostring(xml)
 173         g2 = gerald.Gerald(xml=xml)
 174
 175         # do it all again after extracting from the xml file
 176         self.failUnlessEqual(g.version, g2.version)
 177         self.failUnlessEqual(g.date, g2.date)
 178         self.failUnlessEqual(len(g.lanes.keys()), len(g2.lanes.keys()))
 179         self.failUnlessEqual(len(g.lanes.items()), len(g2.lanes.items()))
 180
 181         # test lane specific parameters from gerald config file
 182         for i in range(1,9):
 183             g_lane = g.lanes[i]
 184             g2_lane = g2.lanes[i]
 185             self.failUnlessEqual(g_lane.analysis, g2_lane.analysis)
 186             self.failUnlessEqual(g_lane.eland_genome, g2_lane.eland_genome)
 187             self.failUnlessEqual(g_lane.read_length, g2_lane.read_length)
 188             self.failUnlessEqual(g_lane.use_bases, g2_lane.use_bases)
 189
 190         # test (some) summary elements
 191         for end in [0,1]:
 192             for i in range(1,9):
 193                 g_summary = g.summary[end][i]
 194                 g2_summary = g2.summary[end][i]
 195                 self.failUnlessEqual(g_summary.cluster, g2_summary.cluster)
 196                 self.failUnlessEqual(g_summary.lane, g2_summary.lane)
 197
 198                 g_eland = g.eland_results
 199                 g2_eland = g2.eland_results
 200                 for lane in g_eland.keys():
 201                     self.failUnlessEqual(g_eland[lane].reads,
 202                                          g2_eland[lane].reads)
 203                     self.failUnlessEqual(len(g_eland[lane].mapped_reads),
 204                                          len(g2_eland[lane].mapped_reads))
 205                     for k in g_eland[lane].mapped_reads.keys():
 206                         self.failUnlessEqual(g_eland[lane].mapped_reads[k],
 207                                              g2_eland[lane].mapped_reads[k])
 208
 209                     self.failUnlessEqual(len(g_eland[lane].match_codes),
 210                                          len(g2_eland[lane].match_codes))
 211                     for k in g_eland[lane].match_codes.keys():
 212                         self.failUnlessEqual(g_eland[lane].match_codes[k],
 213                                              g2_eland[lane].match_codes[k])
 214
 215
 216     def test_eland(self):
 217         hg_map = {'Lambda.fa': 'Lambda.fa'}
 218         for i in range(1,22):
 219           short_name = 'chr%d.fa' % (i,)
 220           long_name = 'hg18/chr%d.fa' % (i,)
 221           hg_map[short_name] = long_name
 222
 223         genome_maps = { 1:hg_map, 2:hg_map, 3:hg_map, 4:hg_map,
 224                         5:hg_map, 6:hg_map, 7:hg_map, 8:hg_map }
 225         eland = gerald.eland(self.gerald_dir, genome_maps=genome_maps)
 226
 227         for i in range(1,9):
 228             lane = eland[i]
 229             self.failUnlessEqual(lane.reads, 4)
 230             self.failUnlessEqual(lane.sample_name, "s")
 231             self.failUnlessEqual(lane.lane_id, i)
 232             self.failUnlessEqual(len(lane.mapped_reads), 15)
 233             self.failUnlessEqual(lane.mapped_reads['hg18/chr5.fa'], 4)
 234             self.failUnlessEqual(lane.match_codes['U0'], 1)
 235             self.failUnlessEqual(lane.match_codes['R0'], 2)
 236             self.failUnlessEqual(lane.match_codes['U1'], 1)
 237             self.failUnlessEqual(lane.match_codes['R1'], 9)
 238             self.failUnlessEqual(lane.match_codes['U2'], 0)
 239             self.failUnlessEqual(lane.match_codes['R2'], 12)
 240             self.failUnlessEqual(lane.match_codes['NM'], 1)
 241             self.failUnlessEqual(lane.match_codes['QC'], 0)
 242
 243         xml = eland.get_elements()
 244         # just make sure that element tree can serialize the tree
 245         xml_str = ElementTree.tostring(xml)
 246         e2 = gerald.ELAND(xml=xml)
 247
 248         for i in range(1,9):
 249             l1 = eland[i]
 250             l2 = e2[i]
 251             self.failUnlessEqual(l1.reads, l2.reads)
 252             self.failUnlessEqual(l1.sample_name, l2.sample_name)
 253             self.failUnlessEqual(l1.lane_id, l2.lane_id)
 254             self.failUnlessEqual(len(l1.mapped_reads), len(l2.mapped_reads))
 255             self.failUnlessEqual(len(l1.mapped_reads), 15)
 256             for k in l1.mapped_reads.keys():
 257                 self.failUnlessEqual(l1.mapped_reads[k],
 258                                      l2.mapped_reads[k])
 259
 260             self.failUnlessEqual(len(l1.match_codes), 9)
 261             self.failUnlessEqual(len(l1.match_codes), len(l2.match_codes))
 262             for k in l1.match_codes.keys():
 263                 self.failUnlessEqual(l1.match_codes[k],
 264                                      l2.match_codes[k])
 265
 266     def test_runfolder(self):
 267         runs = runfolder.get_runs(self.runfolder_dir)
 268
 269         # do we get the flowcell id from the filename?
 270         self.failUnlessEqual(len(runs), 1)
 271         # firecrest's date depends on filename not the create time.
 272         name = 'run_207BTAAXX_2008-04-19.xml'
 273         self.failUnlessEqual(runs[0].name, name)
 274
 275         # do we get the flowcell id from the FlowcellId.xml file
 276         make_flowcell_id(self.runfolder_dir, '207BTAAXY')
 277         runs = runfolder.get_runs(self.runfolder_dir)
 278         self.failUnlessEqual(len(runs), 1)
 279         name = 'run_207BTAAXY_2008-04-19.xml'
 280         self.failUnlessEqual(runs[0].name, name)
 281
 282         r1 = runs[0]
 283         xml = r1.get_elements()
 284         xml_str = ElementTree.tostring(xml)
 285
 286         r2 = runfolder.PipelineRun(xml=xml)
 287         self.failUnlessEqual(r1.name, r2.name)
 288         self.failIfEqual(r2.image_analysis, None)
 289         self.failIfEqual(r2.bustard, None)
 290         self.failIfEqual(r2.gerald, None)
 291
 292
 293 def suite():
 294     return unittest.makeSuite(RunfolderTests,'test')
 295
 296 if __name__ == "__main__":
 297     unittest.main(defaultTest="suite")
 298