htsworkflow/pipelines/test/test_runfolder_ipar130.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                                  '090313_HWI-EAS229_0101_3021JAAXX')
  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     bustard_dir = os.path.join(ipar_dir,
  35                                'Bustard1.3.2_15-03-2008_diane')
  36     os.mkdir(bustard_dir)
  37     make_phasing_params(bustard_dir)
  38
  39     gerald_dir = os.path.join(bustard_dir,
  40                               'GERALD_15-03-2008_diane')
  41     os.mkdir(gerald_dir)
  42     make_gerald_config(gerald_dir)
  43     make_summary_ipar130_htm(gerald_dir)
  44     make_eland_multi(gerald_dir)
  45
  46     if obj is not None:
  47         obj.temp_dir = temp_dir
  48         obj.runfolder_dir = runfolder_dir
  49         obj.data_dir = data_dir
  50         obj.image_analysis_dir = ipar_dir
  51         obj.bustard_dir = bustard_dir
  52         obj.gerald_dir = gerald_dir
  53
  54
  55 class RunfolderTests(unittest.TestCase):
  56     """
  57     Test components of the runfolder processing code
  58     which includes firecrest, bustard, and gerald
  59     """
  60     def setUp(self):
  61         # attaches all the directories to the object passed in
  62         make_runfolder(self)
  63
  64     def tearDown(self):
  65         shutil.rmtree(self.temp_dir)
  66
  67     def test_ipar(self):
  68         """
  69         Construct a firecrest object
  70         """
  71         i = ipar.ipar(self.image_analysis_dir)
  72         self.failUnlessEqual(i.version, '2.01.192.0')
  73         self.failUnlessEqual(i.start, 1)
  74         self.failUnlessEqual(i.stop, 37)
  75
  76         xml = i.get_elements()
  77         # just make sure that element tree can serialize the tree
  78         xml_str = ElementTree.tostring(xml)
  79
  80         i2 = ipar.IPAR(xml=xml)
  81         self.failUnlessEqual(i.version, i2.version)
  82         self.failUnlessEqual(i.start,   i2.start)
  83         self.failUnlessEqual(i.stop,    i2.stop)
  84         self.failUnlessEqual(i.date,    i2.date)
  85         self.failUnlessEqual(i.file_list(), i2.file_list())
  86
  87     def test_bustard(self):
  88         """
  89         construct a bustard object
  90         """
  91         b = bustard.bustard(self.bustard_dir)
  92         self.failUnlessEqual(b.version, '1.3.2')
  93         self.failUnlessEqual(b.date,    date(2008,3,15))
  94         self.failUnlessEqual(b.user,    'diane')
  95         self.failUnlessEqual(len(b.phasing), 8)
  96         self.failUnlessAlmostEqual(b.phasing[8].phasing, 0.0099)
  97
  98         xml = b.get_elements()
  99         b2 = bustard.Bustard(xml=xml)
 100         self.failUnlessEqual(b.version, b2.version)
 101         self.failUnlessEqual(b.date,    b2.date )
 102         self.failUnlessEqual(b.user,    b2.user)
 103         self.failUnlessEqual(len(b.phasing), len(b2.phasing))
 104         for key in b.phasing.keys():
 105             self.failUnlessEqual(b.phasing[key].lane,
 106                                  b2.phasing[key].lane)
 107             self.failUnlessEqual(b.phasing[key].phasing,
 108                                  b2.phasing[key].phasing)
 109             self.failUnlessEqual(b.phasing[key].prephasing,
 110                                  b2.phasing[key].prephasing)
 111
 112     def test_gerald(self):
 113         # need to update gerald and make tests for it
 114         g = gerald.gerald(self.gerald_dir)
 115
 116         self.failUnlessEqual(g.version,
 117             '@(#) Id: GERALD.pl,v 1.68.2.2 2007/06/13 11:08:49 km Exp')
 118         self.failUnlessEqual(g.date, datetime(2008,4,19,19,8,30))
 119         self.failUnlessEqual(len(g.lanes), len(g.lanes.keys()))
 120         self.failUnlessEqual(len(g.lanes), len(g.lanes.items()))
 121
 122
 123         # list of genomes, matches what was defined up in
 124         # make_gerald_config.
 125         # the first None is to offset the genomes list to be 1..9
 126         # instead of pythons default 0..8
 127         genomes = [None, '/g/dm3', '/g/equcab1', '/g/equcab1', '/g/canfam2',
 128                          '/g/hg18', '/g/hg18', '/g/hg18', '/g/hg18', ]
 129
 130         # test lane specific parameters from gerald config file
 131         for i in range(1,9):
 132             cur_lane = g.lanes[i]
 133             self.failUnlessEqual(cur_lane.analysis, 'eland')
 134             self.failUnlessEqual(cur_lane.eland_genome, genomes[i])
 135             self.failUnlessEqual(cur_lane.read_length, '32')
 136             self.failUnlessEqual(cur_lane.use_bases, 'Y'*32)
 137
 138         # I want to be able to use a simple iterator
 139         for l in g.lanes.values():
 140           self.failUnlessEqual(l.analysis, 'eland')
 141           self.failUnlessEqual(l.read_length, '32')
 142           self.failUnlessEqual(l.use_bases, 'Y'*32)
 143
 144         # test data extracted from summary file
 145         clusters = [None,
 146                     (126910, 4300), (165739, 6792),
 147                     (196565, 8216), (153897, 8501),
 148                     (135536, 3908), (154083, 9315),
 149                     (159991, 9292), (198479, 17671),]
 150
 151         self.failUnlessEqual(len(g.summary), 1)
 152         for i in range(1,9):
 153             summary_lane = g.summary[0][i]
 154             self.failUnlessEqual(summary_lane.cluster, clusters[i])
 155             self.failUnlessEqual(summary_lane.lane, i)
 156
 157         xml = g.get_elements()
 158         # just make sure that element tree can serialize the tree
 159         xml_str = ElementTree.tostring(xml)
 160         g2 = gerald.Gerald(xml=xml)
 161
 162         # do it all again after extracting from the xml file
 163         self.failUnlessEqual(g.version, g2.version)
 164         self.failUnlessEqual(g.date, g2.date)
 165         self.failUnlessEqual(len(g.lanes.keys()), len(g2.lanes.keys()))
 166         self.failUnlessEqual(len(g.lanes.items()), len(g2.lanes.items()))
 167
 168         # test lane specific parameters from gerald config file
 169         for i in range(1,9):
 170             g_lane = g.lanes[i]
 171             g2_lane = g2.lanes[i]
 172             self.failUnlessEqual(g_lane.analysis, g2_lane.analysis)
 173             self.failUnlessEqual(g_lane.eland_genome, g2_lane.eland_genome)
 174             self.failUnlessEqual(g_lane.read_length, g2_lane.read_length)
 175             self.failUnlessEqual(g_lane.use_bases, g2_lane.use_bases)
 176
 177         # test (some) summary elements
 178         self.failUnlessEqual(len(g.summary), 1)
 179         for i in range(1,9):
 180             g_summary = g.summary[0][i]
 181             g2_summary = g2.summary[0][i]
 182             self.failUnlessEqual(g_summary.cluster, g2_summary.cluster)
 183             self.failUnlessEqual(g_summary.lane, g2_summary.lane)
 184
 185             g_eland = g.eland_results
 186             g2_eland = g2.eland_results
 187             for lane in g_eland.results[0].keys():
 188                 g_results = g_eland.results[0][lane]
 189                 g2_results = g2_eland.results[0][lane]
 190                 self.failUnlessEqual(g_results.reads,
 191                                      g2_results.reads)
 192                 self.failUnlessEqual(len(g_results.mapped_reads),
 193                                      len(g2_results.mapped_reads))
 194                 for k in g_results.mapped_reads.keys():
 195                     self.failUnlessEqual(g_results.mapped_reads[k],
 196                                          g2_results.mapped_reads[k])
 197
 198                 self.failUnlessEqual(len(g_results.match_codes),
 199                                      len(g2_results.match_codes))
 200                 for k in g_results.match_codes.keys():
 201                     self.failUnlessEqual(g_results.match_codes[k],
 202                                          g2_results.match_codes[k])
 203
 204
 205     def test_eland(self):
 206         hg_map = {'Lambda.fa': 'Lambda.fa'}
 207         for i in range(1,22):
 208           short_name = 'chr%d.fa' % (i,)
 209           long_name = 'hg18/chr%d.fa' % (i,)
 210           hg_map[short_name] = long_name
 211
 212         genome_maps = { 1:hg_map, 2:hg_map, 3:hg_map, 4:hg_map,
 213                         5:hg_map, 6:hg_map, 7:hg_map, 8:hg_map }
 214         eland = gerald.eland(self.gerald_dir, genome_maps=genome_maps)
 215
 216         for i in range(1,9):
 217             lane = eland.results[0][i]
 218             self.failUnlessEqual(lane.reads, 6)
 219             self.failUnlessEqual(lane.sample_name, "s")
 220             self.failUnlessEqual(lane.lane_id, i)
 221             self.failUnlessEqual(len(lane.mapped_reads), 17)
 222             self.failUnlessEqual(lane.mapped_reads['hg18/chr5.fa'], 4)
 223             self.failUnlessEqual(lane.match_codes['U0'], 3)
 224             self.failUnlessEqual(lane.match_codes['R0'], 2)
 225             self.failUnlessEqual(lane.match_codes['U1'], 1)
 226             self.failUnlessEqual(lane.match_codes['R1'], 9)
 227             self.failUnlessEqual(lane.match_codes['U2'], 0)
 228             self.failUnlessEqual(lane.match_codes['R2'], 12)
 229             self.failUnlessEqual(lane.match_codes['NM'], 1)
 230             self.failUnlessEqual(lane.match_codes['QC'], 0)
 231
 232         xml = eland.get_elements()
 233         # just make sure that element tree can serialize the tree
 234         xml_str = ElementTree.tostring(xml)
 235         e2 = gerald.ELAND(xml=xml)
 236
 237         for i in range(1,9):
 238             l1 = eland.results[0][i]
 239             l2 = e2.results[0][i]
 240             self.failUnlessEqual(l1.reads, l2.reads)
 241             self.failUnlessEqual(l1.sample_name, l2.sample_name)
 242             self.failUnlessEqual(l1.lane_id, l2.lane_id)
 243             self.failUnlessEqual(len(l1.mapped_reads), len(l2.mapped_reads))
 244             self.failUnlessEqual(len(l1.mapped_reads), 17)
 245             for k in l1.mapped_reads.keys():
 246                 self.failUnlessEqual(l1.mapped_reads[k],
 247                                      l2.mapped_reads[k])
 248
 249             self.failUnlessEqual(len(l1.match_codes), 9)
 250             self.failUnlessEqual(len(l1.match_codes), len(l2.match_codes))
 251             for k in l1.match_codes.keys():
 252                 self.failUnlessEqual(l1.match_codes[k],
 253                                      l2.match_codes[k])
 254
 255     def test_runfolder(self):
 256         runs = runfolder.get_runs(self.runfolder_dir)
 257
 258         # do we get the flowcell id from the filename?
 259         self.failUnlessEqual(len(runs), 1)
 260         name = 'run_3021JAAXX_%s.xml' % ( date.today().strftime('%Y-%m-%d'),)
 261         self.failUnlessEqual(runs[0].name, name)
 262
 263         # do we get the flowcell id from the FlowcellId.xml file
 264         make_flowcell_id(self.runfolder_dir, '207BTAAXY')
 265         runs = runfolder.get_runs(self.runfolder_dir)
 266         self.failUnlessEqual(len(runs), 1)
 267         name = 'run_207BTAAXY_%s.xml' % ( date.today().strftime('%Y-%m-%d'),)
 268         self.failUnlessEqual(runs[0].name, name)
 269
 270         r1 = runs[0]
 271         xml = r1.get_elements()
 272         xml_str = ElementTree.tostring(xml)
 273
 274         r2 = runfolder.PipelineRun(xml=xml)
 275         self.failUnlessEqual(r1.name, r2.name)
 276         self.failIfEqual(r2.image_analysis, None)
 277         self.failIfEqual(r2.bustard, None)
 278         self.failIfEqual(r2.gerald, None)
 279
 280
 281 def suite():
 282     return unittest.makeSuite(RunfolderTests,'test')
 283
 284 if __name__ == "__main__":
 285     unittest.main(defaultTest="suite")
 286