def test_simple_array(self):
a1 = np.array(list('axzbykcq'))
a2 = np.array(list('12ab34'))
i1, i2 = np_ext.index_intersect(a1, a2)
self.assertEqual(i1.tolist(), [0, 3])
self.assertEqual(i2.tolist(), [2, 3])
i3, i4 = np_ext.index_intersect(a2, a1)
self.assertTrue(np.array_equal(i4, i1))
self.assertTrue(np.array_equal(i3, i2))
def test_simple_array(self):
a1 = np.array(list('axzbykcq'))
a2 = np.array(list('12ab34'))
i1, i2 = np_ext.index_intersect(a1, a2)
self.assertEqual(i1.tolist(), [0, 3])
self.assertEqual(i2.tolist(), [2, 3])
i3, i4 = np_ext.index_intersect(a2, a1)
self.assertTrue(np.array_equal(i4, i1))
self.assertTrue(np.array_equal(i3, i2))
def test_performance(self): size = 1000000 offset = size/10000 a1 = np.array(['V%07d' % i for i in xrange(offset, size)]) a2 = np.array(['V%07d' % i for i in xrange(offset+size)]) t0 = time.time() i1, i2 = np_ext.index_intersect(a1, a2) print print "finished in %.1f s" % (time.time()-t0)
def test_performance(self):
size = 1000000
offset = size / 10000
a1 = np.array(['V%07d' % i for i in xrange(offset, size)])
a2 = np.array(['V%07d' % i for i in xrange(offset + size)])
t0 = time.time()
i1, i2 = np_ext.index_intersect(a1, a2)
print
print "finished in %.1f s" % (time.time() - t0)
def test_record_array(self):
a1 = np.array([('V002', 0), ('V068', 1), ('V129', 0)],
dtype=[('marker_vid', '|S34'), ('allele_flip', '|i1')])
a2 = np.array([('V%03d' % i, 'foo-%03d' % i) for i in xrange(1000)],
dtype=[('vid', '|S34'), ('label', '|S48')])
i1, i2 = np_ext.index_intersect(a1['marker_vid'], a2['vid'])
self.assertEqual(i1.tolist(), [0, 1, 2])
self.assertEqual(i2.tolist(), [2, 68, 129])
self.assertEqual(a1[i1].tolist(), [('V002', 0), ('V068', 1), ('V129', 0)])
self.assertEqual(
a2[i2].tolist(),
[('V002', 'foo-002'), ('V068', 'foo-068'), ('V129', 'foo-129')]
)
def test_record_array(self):
a1 = np.array([('V002', 0), ('V068', 1), ('V129', 0)],
dtype=[('marker_vid', '|S34'), ('allele_flip', '|i1')])
a2 = np.array([('V%03d' % i, 'foo-%03d' % i) for i in xrange(1000)],
dtype=[('vid', '|S34'), ('label', '|S48')])
i1, i2 = np_ext.index_intersect(a1['marker_vid'], a2['vid'])
self.assertEqual(i1.tolist(), [0, 1, 2])
self.assertEqual(i2.tolist(), [2, 68, 129])
self.assertEqual(a1[i1].tolist(), [('V002', 0), ('V068', 1),
('V129', 0)])
self.assertEqual(a2[i2].tolist(), [('V002', 'foo-002'),
('V068', 'foo-068'),
('V129', 'foo-129')])
def load_markers(self, batch_size=1000, additional_fields=None):
"""
Read marker info from the marker set table and store it in the
markers attribute.
If additional_fields is provided, it must be a list of fields from
the marker definition table; in this case, the additional info is
stored in the add_marker_info attribute.
"""
data = self.proxy.gadpt.read_snp_markers_set(self.id, batch_size=batch_size)
self.__set_markers(data)
if additional_fields is not None:
if "vid" not in additional_fields:
additional_fields.append("vid")
recs = self.proxy.get_snp_marker_definitions(col_names=additional_fields,
batch_size=batch_size)
i1, i2 = np_ext.index_intersect(data['marker_vid'], recs['vid'])
recs = recs[i2]
# FIXME: this is not very efficient
by_vid = dict((r['vid'], r) for r in recs)
recs = np.array([by_vid[d['marker_vid']] for d in data], dtype=recs.dtype)
self.__set_add_marker_info(recs)
def intersect(mset1, mset2):
"""
Returns a pair of equal length numpy arrays where corresponding
array elements are the indices of markers, respectively in mset1
and mset2, that align to the same position on the same ref_genome.
.. code-block:: python
ref_genome = 'hg19'
ms1.load_alignments(ref_genome)
ms2.load_alignments(ref_genome)
idx1, idx2 = kb.SNPMarkersSet.intersect(ms1, ms2)
for i1, i2 in it.izip(idx1, idx2):
assert ms1[i].position == ms2[i].position
"""
if not (mset1.has_aligns() and mset2.has_aligns()):
raise ValueError('both mset should be aligned')
if mset1.ref_genome != mset2.ref_genome:
raise ValueError('msets should be aligned to the same ref_genome')
gpos1 = mset1.aligns['global_pos']
gpos2 = mset2.aligns['global_pos']
return np_ext.index_intersect(gpos1, gpos2)
def intersect(mset1, mset2):
"""
Returns a pair of equal length numpy arrays where corresponding
array elements are the indices of markers, respectively in mset1
and mset2, that align to the same position on the same ref_genome.
.. code-block:: python
ref_genome = 'hg19'
ms1.load_alignments(ref_genome)
ms2.load_alignments(ref_genome)
idx1, idx2 = kb.SNPMarkersSet.intersect(ms1, ms2)
for i1, i2 in it.izip(idx1, idx2):
assert ms1[i].position == ms2[i].position
"""
if not (mset1.has_aligns() and mset2.has_aligns()):
raise ValueError('both mset should be aligned')
if mset1.ref_genome != mset2.ref_genome:
raise ValueError('msets should be aligned to the same ref_genome')
gpos1 = mset1.aligns['global_pos']
gpos2 = mset2.aligns['global_pos']
return np_ext.index_intersect(gpos1, gpos2)
def load_markers(self, batch_size=1000, additional_fields=None):
"""
Read marker info from the marker set table and store it in the
markers attribute.
If additional_fields is provided, it must be a list of fields from
the marker definition table; in this case, the additional info is
stored in the add_marker_info attribute.
"""
data = self.proxy.gadpt.read_snp_markers_set(self.id,
batch_size=batch_size)
self.__set_markers(data)
if additional_fields is not None:
if "vid" not in additional_fields:
additional_fields.append("vid")
recs = self.proxy.get_snp_marker_definitions(
col_names=additional_fields, batch_size=batch_size)
i1, i2 = np_ext.index_intersect(data['marker_vid'], recs['vid'])
recs = recs[i2]
# FIXME: this is not very efficient
by_vid = dict((r['vid'], r) for r in recs)
recs = np.array([by_vid[d['marker_vid']] for d in data],
dtype=recs.dtype)
self.__set_add_marker_info(recs)
