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 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                                  '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(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_firecrest(self):
  74         """
  75         Construct a firecrest object
  76         """
  77         f = firecrest.firecrest(self.image_analysis_dir)
  78         self.failUnlessEqual(f.version, '1.9.6')
  79         self.failUnlessEqual(f.start, 1)
  80         self.failUnlessEqual(f.stop, 37)
  81         self.failUnlessEqual(f.user, 'diane')
  82         self.failUnlessEqual(f.date, date(2008,10,20))
  83
  84         xml = f.get_elements()
  85         # just make sure that element tree can serialize the tree
  86         xml_str = ElementTree.tostring(xml)
  87
  88         f2 = firecrest.Firecrest(xml=xml)
  89         self.failUnlessEqual(f.version, f2.version)
  90         self.failUnlessEqual(f.start,   f2.start)
  91         self.failUnlessEqual(f.stop,    f2.stop)
  92         self.failUnlessEqual(f.user,    f2.user)
  93
  94     def test_bustard(self):
  95         """
  96         construct a bustard object
  97         """
  98         b = bustard.bustard(self.bustard_dir)
  99         self.failUnlessEqual(b.version, '1.9.6')
 100         self.failUnlessEqual(b.date,    date(2008,10,20))
 101         self.failUnlessEqual(b.user,    'diane')
 102         self.failUnlessEqual(len(b.phasing), 8)
 103         self.failUnlessAlmostEqual(b.phasing[8].phasing, 0.0099)
 104
 105         xml = b.get_elements()
 106         b2 = bustard.Bustard(xml=xml)
 107         self.failUnlessEqual(b.version, b2.version)
 108         self.failUnlessEqual(b.date,    b2.date )
 109         self.failUnlessEqual(b.user,    b2.user)
 110         self.failUnlessEqual(len(b.phasing), len(b2.phasing))
 111         for key in b.phasing.keys():
 112             self.failUnlessEqual(b.phasing[key].lane,
 113                                  b2.phasing[key].lane)
 114             self.failUnlessEqual(b.phasing[key].phasing,
 115                                  b2.phasing[key].phasing)
 116             self.failUnlessEqual(b.phasing[key].prephasing,
 117                                  b2.phasing[key].prephasing)
 118
 119     def test_gerald(self):
 120         # need to update gerald and make tests for it
 121         g = gerald.gerald(self.gerald_dir)
 122
 123         self.failUnlessEqual(g.version,
 124             '@(#) Id: GERALD.pl,v 1.171 2008/05/19 17:36:14 mzerara Exp')
 125         self.failUnlessEqual(g.date, datetime(2009,2,22,21,15,59))
 126         self.failUnlessEqual(len(g.lanes), len(g.lanes.keys()))
 127         self.failUnlessEqual(len(g.lanes), len(g.lanes.items()))
 128
 129
 130         # list of genomes, matches what was defined up in
 131         # make_gerald_config.
 132         # the first None is to offset the genomes list to be 1..9
 133         # instead of pythons default 0..8
 134         genomes = [None,
 135                    '/g/mm9',
 136                    '/g/mm9',
 137                    '/g/elegans190',
 138                    '/g/arabidopsis01222004',
 139                    '/g/mm9',
 140                    '/g/mm9',
 141                    '/g/mm9',
 142                    '/g/mm9', ]
 143
 144         # test lane specific parameters from gerald config file
 145         for i in range(1,9):
 146             cur_lane = g.lanes[i]
 147             self.failUnlessEqual(cur_lane.analysis, 'eland_extended')
 148             self.failUnlessEqual(cur_lane.eland_genome, genomes[i])
 149             self.failUnlessEqual(cur_lane.read_length, '37')
 150             self.failUnlessEqual(cur_lane.use_bases, 'Y'*37)
 151
 152         # I want to be able to use a simple iterator
 153         for l in g.lanes.values():
 154           self.failUnlessEqual(l.analysis, 'eland_extended')
 155           self.failUnlessEqual(l.read_length, '37')
 156           self.failUnlessEqual(l.use_bases, 'Y'*37)
 157
 158         # raw cluster numbers extracted from summary file
 159         # its the first +/- value in the lane results summary
 160         # section
 161         clusters = [None,
 162                     (190220, 15118), (190560, 14399),
 163                     (187597, 12369), (204142, 16877),
 164                     (247308, 11600), (204298, 15640),
 165                     (202707, 15404), (198075, 14702),]
 166
 167         self.failUnlessEqual(len(g.summary), 1)
 168         for i in range(1,9):
 169             summary_lane = g.summary[0][i]
 170             self.failUnlessEqual(summary_lane.cluster, clusters[i])
 171             self.failUnlessEqual(summary_lane.lane, i)
 172
 173         xml = g.get_elements()
 174         # just make sure that element tree can serialize the tree
 175         xml_str = ElementTree.tostring(xml)
 176         g2 = gerald.Gerald(xml=xml)
 177
 178         # do it all again after extracting from the xml file
 179         self.failUnlessEqual(g.version, g2.version)
 180         self.failUnlessEqual(g.date, g2.date)
 181         self.failUnlessEqual(len(g.lanes.keys()), len(g2.lanes.keys()))
 182         self.failUnlessEqual(len(g.lanes.items()), len(g2.lanes.items()))
 183
 184         # test lane specific parameters from gerald config file
 185         for i in range(1,9):
 186             g_lane = g.lanes[i]
 187             g2_lane = g2.lanes[i]
 188             self.failUnlessEqual(g_lane.analysis, g2_lane.analysis)
 189             self.failUnlessEqual(g_lane.eland_genome, g2_lane.eland_genome)
 190             self.failUnlessEqual(g_lane.read_length, g2_lane.read_length)
 191             self.failUnlessEqual(g_lane.use_bases, g2_lane.use_bases)
 192
 193         self.failUnlessEqual(len(g.summary), 1)
 194         # test (some) summary elements
 195         for i in range(1,9):
 196             g_summary = g.summary[0][i]
 197             g2_summary = g2.summary[0][i]
 198             self.failUnlessEqual(g_summary.cluster, g2_summary.cluster)
 199             self.failUnlessEqual(g_summary.lane, g2_summary.lane)
 200
 201             g_eland = g.eland_results
 202             g2_eland = g2.eland_results
 203             for lane in g_eland.results[0].keys():
 204                 g_results = g_eland.results[0][lane]
 205                 g2_results = g2_eland.results[0][lane]
 206                 self.failUnlessEqual(g_results.reads,
 207                                      g2_results.reads)
 208                 self.failUnlessEqual(len(g_results.mapped_reads),
 209                                      len(g2_results.mapped_reads))
 210                 for k in g_results.mapped_reads.keys():
 211                     self.failUnlessEqual(g_results.mapped_reads[k],
 212                                          g2_results.mapped_reads[k])
 213
 214                 self.failUnlessEqual(len(g_results.match_codes),
 215                                      len(g2_results.match_codes))
 216                 for k in g_results.match_codes.keys():
 217                     self.failUnlessEqual(g_results.match_codes[k],
 218                                          g2_results.match_codes[k])
 219
 220
 221     def test_eland(self):
 222         hg_map = {'Lambda.fa': 'Lambda.fa'}
 223         for i in range(1,22):
 224           short_name = 'chr%d.fa' % (i,)
 225           long_name = 'hg18/chr%d.fa' % (i,)
 226           hg_map[short_name] = long_name
 227
 228         genome_maps = { 1:hg_map, 2:hg_map, 3:hg_map, 4:hg_map,
 229                         5:hg_map, 6:hg_map, 7:hg_map, 8:hg_map }
 230         eland = gerald.eland(self.gerald_dir, genome_maps=genome_maps)
 231
 232         for i in range(1,9):
 233             lane = eland.results[0][i]
 234             self.failUnlessEqual(lane.reads, 6)
 235             self.failUnlessEqual(lane.sample_name, "s")
 236             self.failUnlessEqual(lane.lane_id, i)
 237             self.failUnlessEqual(len(lane.mapped_reads), 17)
 238             self.failUnlessEqual(lane.mapped_reads['hg18/chr5.fa'], 4)
 239             self.failUnlessEqual(lane.match_codes['U0'], 3)
 240             self.failUnlessEqual(lane.match_codes['R0'], 2)
 241             self.failUnlessEqual(lane.match_codes['U1'], 1)
 242             self.failUnlessEqual(lane.match_codes['R1'], 9)
 243             self.failUnlessEqual(lane.match_codes['U2'], 0)
 244             self.failUnlessEqual(lane.match_codes['R2'], 12)
 245             self.failUnlessEqual(lane.match_codes['NM'], 1)
 246             self.failUnlessEqual(lane.match_codes['QC'], 0)
 247
 248         xml = eland.get_elements()
 249         # just make sure that element tree can serialize the tree
 250         xml_str = ElementTree.tostring(xml)
 251         e2 = gerald.ELAND(xml=xml)
 252
 253         for i in range(1,9):
 254             l1 = eland.results[0][i]
 255             l2 = e2.results[0][i]
 256             self.failUnlessEqual(l1.reads, l2.reads)
 257             self.failUnlessEqual(l1.sample_name, l2.sample_name)
 258             self.failUnlessEqual(l1.lane_id, l2.lane_id)
 259             self.failUnlessEqual(len(l1.mapped_reads), len(l2.mapped_reads))
 260             self.failUnlessEqual(len(l1.mapped_reads), 17)
 261             for k in l1.mapped_reads.keys():
 262                 self.failUnlessEqual(l1.mapped_reads[k],
 263                                      l2.mapped_reads[k])
 264
 265             self.failUnlessEqual(len(l1.match_codes), 9)
 266             self.failUnlessEqual(len(l1.match_codes), len(l2.match_codes))
 267             for k in l1.match_codes.keys():
 268                 self.failUnlessEqual(l1.match_codes[k],
 269                                      l2.match_codes[k])
 270
 271     def test_runfolder(self):
 272         runs = runfolder.get_runs(self.runfolder_dir)
 273
 274         # do we get the flowcell id from the filename?
 275         self.failUnlessEqual(len(runs), 1)
 276         name = 'run_30J55AAXX_2009-02-22.xml'
 277         self.failUnlessEqual(runs[0].name, name)
 278
 279         # do we get the flowcell id from the FlowcellId.xml file
 280         make_flowcell_id(self.runfolder_dir, '30J55AAXX')
 281         runs = runfolder.get_runs(self.runfolder_dir)
 282         self.failUnlessEqual(len(runs), 1)
 283         name = 'run_30J55AAXX_2009-02-22.xml'
 284         self.failUnlessEqual(runs[0].name, name)
 285
 286         r1 = runs[0]
 287         xml = r1.get_elements()
 288         xml_str = ElementTree.tostring(xml)
 289
 290         r2 = runfolder.PipelineRun(xml=xml)
 291         self.failUnlessEqual(r1.name, r2.name)
 292         self.failIfEqual(r2.image_analysis, None)
 293         self.failIfEqual(r2.bustard, None)
 294         self.failIfEqual(r2.gerald, None)
 295
 296
 297 def suite():
 298     return unittest.makeSuite(RunfolderTests,'test')
 299
 300 if __name__ == "__main__":
 301     unittest.main(defaultTest="suite")
 302