tests/nlmsa_test.py

   1 import unittest
   2 from testlib import testutil, PygrTestProgram
   3 from pygr import cnestedlist, nlmsa_utils, seqdb, sequence
   4
   5 class NestedList_Test(unittest.TestCase):
   6     "Basic cnestedlist class tests"
   7
   8     def setUp(self):
   9         self.db = cnestedlist.IntervalDB()
  10         ivals = [(0,10,1,-110,-100), (-20,-5,2,300,315)]
  11         self.db.save_tuples(ivals)
  12
  13     def test_query(self):
  14         "NestedList query"
  15         assert self.db.find_overlap_list(0,10) == \
  16                          [(0, 10, 1, -110, -100), (5, 20, 2, -315, -300)]
  17
  18     def test_reverse(self):
  19         "NestedList reverse"
  20         assert self.db.find_overlap_list(-11,-7) == \
  21                          [(-10, 0, 1, 100, 110), (-20, -5, 2, 300, 315)]
  22
  23     def test_filedb(self):
  24         "NestedList filedb"
  25         tempdir  = testutil.TempDir('nlmsa-test')
  26         filename = tempdir.subfile('nlmsa')
  27         self.db.write_binaries(filename)
  28         fdb=cnestedlist.IntervalFileDB(filename)
  29         assert fdb.find_overlap_list(0,10) == \
  30                          [(0, 10, 1, -110, -100), (5, 20, 2, -315, -300)]
  31         assert fdb.find_overlap_list(-11,-7) == \
  32                          [(-10, 0, 1, 100, 110), (-20, -5, 2, 300, 315)]
  33
  34         # fails on windows
  35         #tempdir.remove()  @CTB
  36
  37 class NLMSA_SimpleTests(unittest.TestCase):
  38
  39     def setUp(self):
  40         pass
  41
  42     def test_empty(self):
  43         "NLMSA Empty"
  44         blasthits  = testutil.tempdatafile('blasthits')
  45
  46         msa = cnestedlist.NLMSA(blasthits , 'memory', pairwiseMode=True)
  47         try:
  48             msa.build()
  49             raise AssertionError('failed to trap empty alignment!')
  50         except nlmsa_utils.EmptyAlignmentError:
  51             pass
  52
  53     def test_empty2(self):
  54         "NLMSA Empty 2"
  55         blasthits = testutil.tempdatafile('blasthits2')
  56         msa = cnestedlist.NLMSA(blasthits, mode='w', pairwiseMode=True)
  57         try:
  58             msa.build()
  59             raise AssertionError('failed to trap empty alignment!')
  60         except nlmsa_utils.EmptyAlignmentError:
  61             pass
  62
  63     def test_build(self):
  64         "NLMSA build"
  65
  66         testnlmsa = testutil.tempdatafile('testnlmsa')
  67         msa = cnestedlist.NLMSA(testnlmsa, mode='w', pairwiseMode=True,
  68                                 bidirectional=False)
  69         # @CTB should there be something else here?  What is this testing?
  70
  71     def test_lpo_query(self):
  72         s1=sequence.Sequence('aaaa','s1')
  73         s2=sequence.Sequence('bbbb','s2')
  74         msa = cnestedlist.NLMSA(mode='memory')
  75         msa[0:4]+=s1
  76         msa[0:4]+=s2
  77         msa.build()
  78         msaSlice = msa[0:4]
  79         assert len(msaSlice) == 2
  80         l = [t[0:2] for t in msaSlice.edges()]
  81         l.sort()
  82         correct = [(slice(0,4),s1),(slice(0,4),s2)]
  83         correct.sort()
  84         assert l == correct
  85
  86 class NLMSA_Test(unittest.TestCase):
  87     def setUp(self):
  88         s = sequence.Sequence('ATGGACAGAGATGACAGATGAC', 'a')
  89         s2 = sequence.Sequence('ATGGGAGCAGCATGACAGATGAC', 'b')
  90
  91         # make a non-empty NLMSA
  92         nlmsa = cnestedlist.NLMSA('foo', mode='memory', pairwiseMode=True)
  93         nlmsa += s
  94         nlmsa[s] += s2
  95         nlmsa.build()
  96
  97         self.s = s
  98         self.s2 = s2
  99         self.nlmsa = nlmsa
 100
 101     def test_iter(self):
 102         "Iteration of NLMSA objects should return reasonable error."
 103
 104         # try iterating over it
 105         try:
 106             for x in self.nlmsa:
 107                 break                   # should fail before this
 108
 109             assert 0, "should not be able to iterate over NLMSA"
 110         except NotImplementedError:
 111             pass
 112
 113     def test_slice_repr(self):
 114         "Ask for an informative __repr__ on NLMSASlice objects"
 115
 116         slice = self.nlmsa[self.s]
 117         r = repr(slice)
 118         assert 'seq=a' in r
 119
 120         slice = self.nlmsa[self.s2]
 121         r = repr(slice)
 122         assert 'seq=b' in r
 123
 124 class NLMSA_BuildWithAlignedIntervals_Test(unittest.TestCase):
 125     def setUp(self):
 126         seqdb_name = testutil.datafile('alignments.fa')
 127         self.db = seqdb.SequenceFileDB(seqdb_name)
 128
 129     def _check_results(self, n):
 130         db = self.db
 131
 132         a, b, c = db['a'], db['b'], db['c']
 133
 134         ival = a[0:8]
 135         (result,) = n[ival].keys()
 136         assert result == b[0:8]
 137
 138         ival = a[12:20]
 139         (result,) = n[ival].keys()
 140         assert result == c[0:8]
 141
 142         l = list(n[a].keys())
 143         l.sort()
 144         assert b[0:8] in l
 145         assert c[0:8] in l
 146
 147     def test_simple(self):
 148         # first set of intervals
 149         ivals = [(('a', 0, 8, 1), ('b', 0, 8, 1),),
 150                  (('a', 12, 20, 1), ('c', 0, 8, 1)),]
 151
 152         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True)
 153
 154         alignedIvalsAttrs = dict(id=0, start=1, stop=2, idDest=0, startDest=1,
 155                                  stopDest=2, ori=3, oriDest=3)
 156         cti = nlmsa_utils.CoordsToIntervals(self.db, self.db,
 157                                             alignedIvalsAttrs)
 158         n.add_aligned_intervals(cti(ivals))
 159         n.build()
 160
 161         self._check_results(n)
 162
 163     def test_simple_no_ori(self):
 164         # first set of intervals
 165         ivals = [(('a', 0, 8,), ('b', 0, 8,),),
 166                  (('a', 12, 20,), ('c', 0, 8,)),]
 167
 168         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True)
 169
 170         alignedIvalsAttrs = dict(id=0, start=1, stop=2, idDest=0, startDest=1,
 171                                  stopDest=2)
 172         cti = nlmsa_utils.CoordsToIntervals(self.db, self.db,
 173                                             alignedIvalsAttrs)
 174         n.add_aligned_intervals(cti(ivals))
 175         n.build()
 176
 177         self._check_results(n)
 178
 179     def test_attr(self):
 180         class Bag(object):
 181             def __init__(self, **kw):
 182                 self.__dict__.update(kw)
 183
 184         # first set of intervals
 185         db = self.db
 186         a, b, c = db['a'], db['b'], db['c']
 187
 188         src_ival1 = Bag(id='a', start=0, stop=8, ori=1)
 189         dst_ival1 = Bag(id='b', start=0, stop=8, ori=1)
 190
 191         src_ival2 = Bag(id='a', start=12, stop=20, ori=1)
 192         dst_ival2 = Bag(id='c', start=0, stop=8, ori=1)
 193
 194         ivals = [(src_ival1, dst_ival1), (src_ival2, dst_ival2)]
 195
 196         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True)
 197
 198         cti = nlmsa_utils.CoordsToIntervals(self.db, self.db)
 199         n.add_aligned_intervals(cti(ivals))
 200         n.build()
 201
 202         self._check_results(n)
 203
 204     def test_single_ival_attr(self):
 205         class Bag(object):
 206             def __init__(self, **kw):
 207                 self.__dict__.update(kw)
 208
 209         # first set of intervals
 210         db = self.db
 211         a, b, c = db['a'], db['b'], db['c']
 212
 213         ival1 = Bag(id='a', start=0, stop=8, ori=1,
 214                     idDest='b', startDest=0, stopDest=8, stopOri=1)
 215         ival2 = Bag(id='a', start=12, stop=20, ori=1,
 216                     idDest='c', startDest=0, stopDest=8, oriDest=1)
 217
 218         ivals = [ival1, ival2]
 219
 220         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True)
 221
 222         cti = nlmsa_utils.CoordsToIntervals(self.db, self.db, {})
 223         n.add_aligned_intervals(cti(ivals))
 224         n.build()
 225
 226         self._check_results(n)
 227
 228
 229     def test_no_seqDict_args(self):
 230         class Bag(object):
 231             def __init__(self, **kw):
 232                 self.__dict__.update(kw)
 233
 234         # first set of intervals
 235         db = self.db
 236
 237         src_ival1 = Bag(id='a', start=0, stop=8, ori=1)
 238         dst_ival1 = Bag(id='b', start=0, stop=8, ori=1)
 239
 240         src_ival2 = Bag(id='a', start=12, stop=20, ori=1)
 241         dst_ival2 = Bag(id='c', start=0, stop=8, ori=1)
 242
 243         ivals = [(src_ival1, dst_ival1), (src_ival2, dst_ival2)]
 244
 245         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True,
 246                               seqDict=db)
 247
 248         cti = nlmsa_utils.CoordsToIntervals(self.db)
 249         n.add_aligned_intervals(cti(ivals))
 250         n.build()
 251
 252 if __name__ == '__main__':
 253     PygrTestProgram(verbosity=2)