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 import unittest
   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.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_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(unittest.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.version, '2.01.192.0')
  79         self.failUnlessEqual(i.start, 1)
  80         self.failUnlessEqual(i.stop, 37)
  81
  82         xml = i.get_elements()
  83         # just make sure that element tree can serialize the tree
  84         xml_str = ElementTree.tostring(xml)
  85
  86         i2 = ipar.IPAR(xml=xml)
  87         self.failUnlessEqual(i.version, i2.version)
  88         self.failUnlessEqual(i.start,   i2.start)
  89         self.failUnlessEqual(i.stop,    i2.stop)
  90         self.failUnlessEqual(i.date,    i2.date)
  91         self.failUnlessEqual(i.file_list(), i2.file_list())
  92
  93     def test_bustard(self):
  94         """
  95         construct a bustard object
  96         """
  97         b = bustard.bustard(self.bustard_dir)
  98         self.failUnlessEqual(b.version, '1.8.28')
  99         self.failUnlessEqual(b.date,    date(2008,4,12))
 100         self.failUnlessEqual(b.user,    'diane')
 101         self.failUnlessEqual(len(b.phasing), 8)
 102         self.failUnlessAlmostEqual(b.phasing[8].phasing, 0.0099)
 103
 104         xml = b.get_elements()
 105         b2 = bustard.Bustard(xml=xml)
 106         self.failUnlessEqual(b.version, b2.version)
 107         self.failUnlessEqual(b.date,    b2.date )
 108         self.failUnlessEqual(b.user,    b2.user)
 109         self.failUnlessEqual(len(b.phasing), len(b2.phasing))
 110         for key in b.phasing.keys():
 111             self.failUnlessEqual(b.phasing[key].lane,
 112                                  b2.phasing[key].lane)
 113             self.failUnlessEqual(b.phasing[key].phasing,
 114                                  b2.phasing[key].phasing)
 115             self.failUnlessEqual(b.phasing[key].prephasing,
 116                                  b2.phasing[key].prephasing)
 117
 118     def test_gerald(self):
 119         # need to update gerald and make tests for it
 120         g = gerald.gerald(self.gerald_dir)
 121
 122         self.failUnlessEqual(g.version, '1.171')
 123         self.failUnlessEqual(g.date, datetime(2009,2,22,21,15,59))
 124         self.failUnlessEqual(len(g.lanes), len(g.lanes.keys()))
 125         self.failUnlessEqual(len(g.lanes), len(g.lanes.items()))
 126
 127
 128         # list of genomes, matches what was defined up in
 129         # make_gerald_config.
 130         # the first None is to offset the genomes list to be 1..9
 131         # instead of pythons default 0..8
 132         genomes = [None,
 133                    '/g/mm9',
 134                    '/g/mm9',
 135                    '/g/elegans190',
 136                    '/g/arabidopsis01222004',
 137                    '/g/mm9',
 138                    '/g/mm9',
 139                    '/g/mm9',
 140                    '/g/mm9', ]
 141
 142         # test lane specific parameters from gerald config file
 143         for i in range(1,9):
 144             cur_lane = g.lanes[i]
 145             self.failUnlessEqual(cur_lane.analysis, 'eland_extended')
 146             self.failUnlessEqual(cur_lane.eland_genome, genomes[i])
 147             self.failUnlessEqual(cur_lane.read_length, '37')
 148             self.failUnlessEqual(cur_lane.use_bases, 'Y'*37)
 149
 150         # I want to be able to use a simple iterator
 151         for l in g.lanes.values():
 152           self.failUnlessEqual(l.analysis, 'eland_extended')
 153           self.failUnlessEqual(l.read_length, '37')
 154           self.failUnlessEqual(l.use_bases, 'Y'*37)
 155
 156         # test data extracted from summary file
 157         clusters = [None,
 158                     (96483, 9074), (133738, 7938),
 159                     (152142, 10002), (15784, 2162),
 160                     (119735, 8465), (152177, 8146),
 161                     (84649, 7325), (54622, 4812),]
 162
 163         self.failUnlessEqual(len(g.summary), 1)
 164         for i in range(1,9):
 165             summary_lane = g.summary[0][i]
 166             self.failUnlessEqual(summary_lane.cluster, clusters[i])
 167             self.failUnlessEqual(summary_lane.lane, i)
 168
 169         xml = g.get_elements()
 170         # just make sure that element tree can serialize the tree
 171         xml_str = ElementTree.tostring(xml)
 172         g2 = gerald.Gerald(xml=xml)
 173
 174         # do it all again after extracting from the xml file
 175         self.failUnlessEqual(g.version, g2.version)
 176         self.failUnlessEqual(g.date, g2.date)
 177         self.failUnlessEqual(len(g.lanes.keys()), len(g2.lanes.keys()))
 178         self.failUnlessEqual(len(g.lanes.items()), len(g2.lanes.items()))
 179
 180         # test lane specific parameters from gerald config file
 181         for i in range(1,9):
 182             g_lane = g.lanes[i]
 183             g2_lane = g2.lanes[i]
 184             self.failUnlessEqual(g_lane.analysis, g2_lane.analysis)
 185             self.failUnlessEqual(g_lane.eland_genome, g2_lane.eland_genome)
 186             self.failUnlessEqual(g_lane.read_length, g2_lane.read_length)
 187             self.failUnlessEqual(g_lane.use_bases, g2_lane.use_bases)
 188
 189         # test (some) summary elements
 190         self.failUnlessEqual(len(g.summary), 1)
 191         for i in range(1,9):
 192             g_summary = g.summary[0][i]
 193             g2_summary = g2.summary[0][i]
 194             self.failUnlessEqual(g_summary.cluster, g2_summary.cluster)
 195             self.failUnlessEqual(g_summary.lane, g2_summary.lane)
 196
 197             g_eland = g.eland_results
 198             g2_eland = g2.eland_results
 199             for lane in g_eland.results[0].keys():
 200                 g_results = g_eland.results[0][lane]
 201                 g2_results = g2_eland.results[0][lane]
 202                 self.failUnlessEqual(g_results.reads,
 203                                      g2_results.reads)
 204                 self.failUnlessEqual(len(g_results.mapped_reads),
 205                                      len(g2_results.mapped_reads))
 206                 for k in g_results.mapped_reads.keys():
 207                     self.failUnlessEqual(g_results.mapped_reads[k],
 208                                          g2_results.mapped_reads[k])
 209
 210                 self.failUnlessEqual(len(g_results.match_codes),
 211                                      len(g2_results.match_codes))
 212                 for k in g_results.match_codes.keys():
 213                     self.failUnlessEqual(g_results.match_codes[k],
 214                                          g2_results.match_codes[k])
 215
 216
 217     def test_eland(self):
 218         hg_map = {'Lambda.fa': 'Lambda.fa'}
 219         for i in range(1,22):
 220           short_name = 'chr%d.fa' % (i,)
 221           long_name = 'hg18/chr%d.fa' % (i,)
 222           hg_map[short_name] = long_name
 223
 224         genome_maps = { 1:hg_map, 2:hg_map, 3:hg_map, 4:hg_map,
 225                         5:hg_map, 6:hg_map, 7:hg_map, 8:hg_map }
 226         eland = gerald.eland(self.gerald_dir, genome_maps=genome_maps)
 227
 228         for i in range(1,9):
 229             lane = eland.results[0][i]
 230             self.failUnlessEqual(lane.reads, 6)
 231             self.failUnlessEqual(lane.sample_name, "s")
 232             self.failUnlessEqual(lane.lane_id, i)
 233             self.failUnlessEqual(len(lane.mapped_reads), 17)
 234             self.failUnlessEqual(lane.mapped_reads['hg18/chr5.fa'], 4)
 235             self.failUnlessEqual(lane.mapped_reads['spike.fa/sample1'], 1)
 236             self.failUnlessEqual(lane.mapped_reads['spike.fa/sample2'], 1)
 237             self.failUnlessEqual(lane.match_codes['U0'], 3)
 238             self.failUnlessEqual(lane.match_codes['R0'], 2)
 239             self.failUnlessEqual(lane.match_codes['U1'], 1)
 240             self.failUnlessEqual(lane.match_codes['R1'], 9)
 241             self.failUnlessEqual(lane.match_codes['U2'], 0)
 242             self.failUnlessEqual(lane.match_codes['R2'], 12)
 243             self.failUnlessEqual(lane.match_codes['NM'], 1)
 244             self.failUnlessEqual(lane.match_codes['QC'], 0)
 245
 246         xml = eland.get_elements()
 247         # just make sure that element tree can serialize the tree
 248         xml_str = ElementTree.tostring(xml)
 249         e2 = gerald.ELAND(xml=xml)
 250
 251         for i in range(1,9):
 252             l1 = eland.results[0][i]
 253             l2 = e2.results[0][i]
 254             self.failUnlessEqual(l1.reads, l2.reads)
 255             self.failUnlessEqual(l1.sample_name, l2.sample_name)
 256             self.failUnlessEqual(l1.lane_id, l2.lane_id)
 257             self.failUnlessEqual(len(l1.mapped_reads), len(l2.mapped_reads))
 258             self.failUnlessEqual(len(l1.mapped_reads), 17)
 259             for k in l1.mapped_reads.keys():
 260                 self.failUnlessEqual(l1.mapped_reads[k],
 261                                      l2.mapped_reads[k])
 262
 263             self.failUnlessEqual(len(l1.match_codes), 9)
 264             self.failUnlessEqual(len(l1.match_codes), len(l2.match_codes))
 265             for k in l1.match_codes.keys():
 266                 self.failUnlessEqual(l1.match_codes[k],
 267                                      l2.match_codes[k])
 268
 269     def test_runfolder(self):
 270         runs = runfolder.get_runs(self.runfolder_dir)
 271
 272         # do we get the flowcell id from the filename?
 273         self.failUnlessEqual(len(runs), 1)
 274         name = 'run_207BTAAXX_%s.xml' % ( date.today().strftime('%Y-%m-%d'),)
 275         self.failUnlessEqual(runs[0].name, name)
 276
 277         # do we get the flowcell id from the FlowcellId.xml file
 278         make_flowcell_id(self.runfolder_dir, '207BTAAXY')
 279         runs = runfolder.get_runs(self.runfolder_dir)
 280         self.failUnlessEqual(len(runs), 1)
 281         name = 'run_207BTAAXY_%s.xml' % ( date.today().strftime('%Y-%m-%d'),)
 282         self.failUnlessEqual(runs[0].name, name)
 283
 284         r1 = runs[0]
 285         xml = r1.get_elements()
 286         xml_str = ElementTree.tostring(xml)
 287
 288         r2 = runfolder.PipelineRun(xml=xml)
 289         self.failUnlessEqual(r1.name, r2.name)
 290         self.failIfEqual(r2.image_analysis, None)
 291         self.failIfEqual(r2.bustard, None)
 292         self.failIfEqual(r2.gerald, None)
 293
 294
 295 def suite():
 296     return unittest.makeSuite(RunfolderTests,'test')
 297
 298 if __name__ == "__main__":
 299     unittest.main(defaultTest="suite")
 300