Library view now shows when a library has archived data.
[htsworkflow.git] / trunk / 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     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_paired_htm(gerald_dir)
49     make_eland_multi(gerald_dir, paired=True)
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_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, 152)
81         self.failUnlessEqual(f.user, 'diane')
82         # As of 2008-12-8, the date was being set in 
83         # simulate_runfolder.make_firecrest_dir
84         self.failUnlessEqual(f.date, date(2008,4,12))
85
86         xml = f.get_elements()
87         # just make sure that element tree can serialize the tree
88         xml_str = ElementTree.tostring(xml)
89
90         f2 = firecrest.Firecrest(xml=xml)
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.version, '1.8.28')
102         self.failUnlessEqual(b.date,    date(2008,4,12))
103         self.failUnlessEqual(b.user,    'diane')
104         self.failUnlessEqual(len(b.phasing), 8)
105         self.failUnlessAlmostEqual(b.phasing[8].phasing, 0.0099)
106
107         xml = b.get_elements()
108         b2 = bustard.Bustard(xml=xml)
109         self.failUnlessEqual(b.version, b2.version)
110         self.failUnlessEqual(b.date,    b2.date )
111         self.failUnlessEqual(b.user,    b2.user)
112         self.failUnlessEqual(len(b.phasing), len(b2.phasing))
113         for key in b.phasing.keys():
114             self.failUnlessEqual(b.phasing[key].lane,
115                                  b2.phasing[key].lane)
116             self.failUnlessEqual(b.phasing[key].phasing,
117                                  b2.phasing[key].phasing)
118             self.failUnlessEqual(b.phasing[key].prephasing,
119                                  b2.phasing[key].prephasing)
120
121     def test_gerald(self):
122         # need to update gerald and make tests for it
123         g = gerald.gerald(self.gerald_dir)
124
125         self.failUnlessEqual(g.version,
126             '@(#) Id: GERALD.pl,v 1.171 2008/05/19 17:36:14 mzerara Exp')
127         self.failUnlessEqual(g.date, datetime(2009,2,22,21,15,59))
128         self.failUnlessEqual(len(g.lanes), len(g.lanes.keys()))
129         self.failUnlessEqual(len(g.lanes), len(g.lanes.items()))
130
131
132         # list of genomes, matches what was defined up in
133         # make_gerald_config.
134         # the first None is to offset the genomes list to be 1..9
135         # instead of pythons default 0..8
136         genomes = [None, 
137                    '/g/mm9', 
138                    '/g/mm9', 
139                    '/g/elegans190', 
140                    '/g/arabidopsis01222004',
141                    '/g/mm9', 
142                    '/g/mm9', 
143                    '/g/mm9', 
144                    '/g/mm9', ]
145
146         # test lane specific parameters from gerald config file
147         for i in range(1,9):
148             cur_lane = g.lanes[i]
149             self.failUnlessEqual(cur_lane.analysis, 'eland_extended')
150             self.failUnlessEqual(cur_lane.eland_genome, genomes[i])
151             self.failUnlessEqual(cur_lane.read_length, '37')
152             self.failUnlessEqual(cur_lane.use_bases, 'Y'*37)
153
154         # I want to be able to use a simple iterator
155         for l in g.lanes.values():
156           self.failUnlessEqual(l.analysis, 'eland_extended')
157           self.failUnlessEqual(l.read_length, '37')
158           self.failUnlessEqual(l.use_bases, 'Y'*37)
159
160         # test data extracted from summary file
161         clusters = [[None,
162                     (103646, 4515), (106678, 4652),
163                     (84583, 5963), (68813, 4782),
164                     (104854, 4664), (43555, 1632),
165                     (54265, 1588), (64363, 2697),],
166                     [None,
167                     (103647, 4516), (106679, 4653),
168                     (84584, 5964), (68814, 4783),
169                     (104855, 4665), (43556, 1633),
170                     (54266, 1589), (64364, 2698),],]
171
172         for end in [0,1]:
173             for lane in range(1,9):
174                 summary_lane = g.summary[end][lane]
175                 self.failUnlessEqual(summary_lane.cluster, clusters[end][lane])
176                 self.failUnlessEqual(summary_lane.lane, lane)
177
178         xml = g.get_elements()
179         # just make sure that element tree can serialize the tree
180         xml_str = ElementTree.tostring(xml)
181         g2 = gerald.Gerald(xml=xml)
182
183         # do it all again after extracting from the xml file
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         # test (some) summary elements
199         for end in [0,1]:
200             for i in range(1,9):
201                 g_summary = g.summary[end][i]
202                 g2_summary = g2.summary[end][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 lane in g_eland.results[end].keys():
209                     g_results = g_eland.results[end][lane]
210                     g2_results = g_eland.results[end][lane]
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         # check first end
238         for i in range(1,9):
239             lane = eland.results[0][i]
240             self.failUnlessEqual(lane.reads, 6)
241             self.failUnlessEqual(lane.sample_name, "s")
242             self.failUnlessEqual(lane.lane_id, i)
243             self.failUnlessEqual(len(lane.mapped_reads), 17)
244             self.failUnlessEqual(lane.mapped_reads['hg18/chr5.fa'], 4)
245             self.failUnlessEqual(lane.match_codes['U0'], 3)
246             self.failUnlessEqual(lane.match_codes['R0'], 2)
247             self.failUnlessEqual(lane.match_codes['U1'], 1)
248             self.failUnlessEqual(lane.match_codes['R1'], 9)
249             self.failUnlessEqual(lane.match_codes['U2'], 0)
250             self.failUnlessEqual(lane.match_codes['R2'], 12)
251             self.failUnlessEqual(lane.match_codes['NM'], 1)
252             self.failUnlessEqual(lane.match_codes['QC'], 0)
253
254         # check second end
255         for i in range(1,9):
256             lane = eland.results[1][i]
257             self.failUnlessEqual(lane.reads, 7)
258             self.failUnlessEqual(lane.sample_name, "s")
259             self.failUnlessEqual(lane.lane_id, i)
260             self.failUnlessEqual(len(lane.mapped_reads), 17)
261             self.failUnlessEqual(lane.mapped_reads['hg18/chr5.fa'], 4)
262             self.failUnlessEqual(lane.match_codes['U0'], 3)
263             self.failUnlessEqual(lane.match_codes['R0'], 2)
264             self.failUnlessEqual(lane.match_codes['U1'], 1)
265             self.failUnlessEqual(lane.match_codes['R1'], 9)
266             self.failUnlessEqual(lane.match_codes['U2'], 0)
267             self.failUnlessEqual(lane.match_codes['R2'], 12)
268             self.failUnlessEqual(lane.match_codes['NM'], 1)
269             self.failUnlessEqual(lane.match_codes['QC'], 1)
270
271         xml = eland.get_elements()
272         # just make sure that element tree can serialize the tree
273         xml_str = ElementTree.tostring(xml)
274         e2 = gerald.ELAND(xml=xml)
275
276         for end in [0, 1]:
277             for i in range(1,9):
278                 l1 = eland.results[end][i]
279                 l2 = e2.results[end][i]
280                 self.failUnlessEqual(l1.reads, l2.reads)
281                 self.failUnlessEqual(l1.sample_name, l2.sample_name)
282                 self.failUnlessEqual(l1.lane_id, l2.lane_id)
283                 self.failUnlessEqual(len(l1.mapped_reads), len(l2.mapped_reads))
284                 self.failUnlessEqual(len(l1.mapped_reads), 17)
285                 for k in l1.mapped_reads.keys():
286                     self.failUnlessEqual(l1.mapped_reads[k],
287                                          l2.mapped_reads[k])
288
289                 self.failUnlessEqual(len(l1.match_codes), 9)
290                 self.failUnlessEqual(len(l1.match_codes), len(l2.match_codes))
291                 for k in l1.match_codes.keys():
292                     self.failUnlessEqual(l1.match_codes[k],
293                                          l2.match_codes[k])
294
295     def test_runfolder(self):
296         runs = runfolder.get_runs(self.runfolder_dir)
297
298         # do we get the flowcell id from the filename?
299         self.failUnlessEqual(len(runs), 1)
300         # firecrest's date depends on filename not the create time.
301         name = 'run_207BTAAXX_2009-02-22.xml'
302         self.failUnlessEqual(runs[0].name, name)
303
304         # do we get the flowcell id from the FlowcellId.xml file
305         make_flowcell_id(self.runfolder_dir, '207BTAAXY')
306         runs = runfolder.get_runs(self.runfolder_dir)
307         self.failUnlessEqual(len(runs), 1)
308         name = 'run_207BTAAXY_2009-02-22.xml'
309         self.failUnlessEqual(runs[0].name, name)
310
311         r1 = runs[0]
312         xml = r1.get_elements()
313         xml_str = ElementTree.tostring(xml)
314
315         r2 = runfolder.PipelineRun(xml=xml)
316         self.failUnlessEqual(r1.name, r2.name)
317         self.failIfEqual(r2.image_analysis, None)
318         self.failIfEqual(r2.bustard, None)
319         self.failIfEqual(r2.gerald, None)
320
321
322 def suite():
323     return unittest.makeSuite(RunfolderTests,'test')
324
325 if __name__ == "__main__":
326     unittest.main(defaultTest="suite")
327