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 class NLMSA_Test(unittest.TestCase):
  72     def setUp(self):
  73         s = sequence.Sequence('ATGGACAGAGATGACAGATGAC', 'a')
  74         s2 = sequence.Sequence('ATGGGAGCAGCATGACAGATGAC', 'b')
  75
  76         # make a non-empty NLMSA
  77         nlmsa = cnestedlist.NLMSA('foo', mode='memory', pairwiseMode=True)
  78         nlmsa += s
  79         nlmsa[s] += s2
  80         nlmsa.build()
  81
  82         self.s = s
  83         self.s2 = s2
  84         self.nlmsa = nlmsa
  85
  86     def test_iter(self):
  87         "Iteration of NLMSA objects should return reasonable error."
  88
  89         # try iterating over it
  90         try:
  91             for x in self.nlmsa:
  92                 break                   # should fail before this
  93
  94             assert 0, "should not be able to iterate over NLMSA"
  95         except NotImplementedError:
  96             pass
  97
  98     def test_slice_repr(self):
  99         "Ask for an informative __repr__ on NLMSASlice objects"
 100
 101         slice = self.nlmsa[self.s]
 102         r = repr(slice)
 103         assert 'seq=a' in r
 104
 105         slice = self.nlmsa[self.s2]
 106         r = repr(slice)
 107         assert 'seq=b' in r
 108
 109 class NLMSA_BuildWithAlignedIntervals_Test(unittest.TestCase):
 110     def setUp(self):
 111         seqdb_name = testutil.datafile('alignments.fa')
 112         self.db = seqdb.SequenceFileDB(seqdb_name)
 113
 114     def _check_results(self, n):
 115         db = self.db
 116
 117         a, b, c = db['a'], db['b'], db['c']
 118
 119         ival = a[0:8]
 120         (result,) = n[ival].keys()
 121         assert result == b[0:8]
 122
 123         ival = a[12:20]
 124         (result,) = n[ival].keys()
 125         assert result == c[0:8]
 126
 127         l = list(n[a].keys())
 128         l.sort()
 129         assert b[0:8] in l
 130         assert c[0:8] in l
 131
 132     def test_simple(self):
 133         # first set of intervals
 134         ivals = [(('a', 0, 8, 1), ('b', 0, 8, 1),),
 135                  (('a', 12, 20, 1), ('c', 0, 8, 1)),]
 136
 137         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True)
 138
 139         alignedIvalsAttrs = dict(id=0, start=1, stop=2, idDest=0, startDest=1,
 140                                  stopDest=2, ori=3, oriDest=3)
 141         cti = nlmsa_utils.CoordsToIntervals(self.db, self.db,
 142                                             alignedIvalsAttrs)
 143         n.add_aligned_intervals(cti(ivals))
 144         n.build()
 145
 146         self._check_results(n)
 147
 148     def test_simple_no_ori(self):
 149         # first set of intervals
 150         ivals = [(('a', 0, 8,), ('b', 0, 8,),),
 151                  (('a', 12, 20,), ('c', 0, 8,)),]
 152
 153         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True)
 154
 155         alignedIvalsAttrs = dict(id=0, start=1, stop=2, idDest=0, startDest=1,
 156                                  stopDest=2)
 157         cti = nlmsa_utils.CoordsToIntervals(self.db, self.db,
 158                                             alignedIvalsAttrs)
 159         n.add_aligned_intervals(cti(ivals))
 160         n.build()
 161
 162         self._check_results(n)
 163
 164     def test_attr(self):
 165         class Bag(object):
 166             def __init__(self, **kw):
 167                 self.__dict__.update(kw)
 168
 169         # first set of intervals
 170         db = self.db
 171         a, b, c = db['a'], db['b'], db['c']
 172
 173         src_ival1 = Bag(id='a', start=0, stop=8, ori=1)
 174         dst_ival1 = Bag(id='b', start=0, stop=8, ori=1)
 175
 176         src_ival2 = Bag(id='a', start=12, stop=20, ori=1)
 177         dst_ival2 = Bag(id='c', start=0, stop=8, ori=1)
 178
 179         ivals = [(src_ival1, dst_ival1), (src_ival2, dst_ival2)]
 180
 181         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True)
 182
 183         cti = nlmsa_utils.CoordsToIntervals(self.db, self.db)
 184         n.add_aligned_intervals(cti(ivals))
 185         n.build()
 186
 187         self._check_results(n)
 188
 189     def test_single_ival_attr(self):
 190         class Bag(object):
 191             def __init__(self, **kw):
 192                 self.__dict__.update(kw)
 193
 194         # first set of intervals
 195         db = self.db
 196         a, b, c = db['a'], db['b'], db['c']
 197
 198         ival1 = Bag(id='a', start=0, stop=8, ori=1,
 199                     idDest='b', startDest=0, stopDest=8, stopOri=1)
 200         ival2 = Bag(id='a', start=12, stop=20, ori=1,
 201                     idDest='c', startDest=0, stopDest=8, oriDest=1)
 202
 203         ivals = [ival1, ival2]
 204
 205         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True)
 206
 207         cti = nlmsa_utils.CoordsToIntervals(self.db, self.db, {})
 208         n.add_aligned_intervals(cti(ivals))
 209         n.build()
 210
 211         self._check_results(n)
 212
 213
 214     def test_no_seqDict_args(self):
 215         class Bag(object):
 216             def __init__(self, **kw):
 217                 self.__dict__.update(kw)
 218
 219         # first set of intervals
 220         db = self.db
 221
 222         src_ival1 = Bag(id='a', start=0, stop=8, ori=1)
 223         dst_ival1 = Bag(id='b', start=0, stop=8, ori=1)
 224
 225         src_ival2 = Bag(id='a', start=12, stop=20, ori=1)
 226         dst_ival2 = Bag(id='c', start=0, stop=8, ori=1)
 227
 228         ivals = [(src_ival1, dst_ival1), (src_ival2, dst_ival2)]
 229
 230         n = cnestedlist.NLMSA('test', mode='memory', pairwiseMode=True,
 231                               seqDict=db)
 232
 233         cti = nlmsa_utils.CoordsToIntervals(self.db)
 234         n.add_aligned_intervals(cti(ivals))
 235         n.build()
 236
 237 if __name__ == '__main__':
 238     PygrTestProgram(verbosity=2)