def nw_metric(self, s1, s2):
"""May or may not produce a true metric. Details in:
E. Halpering, J. Buhler, R. Karp, R. Krauthgamer, and B. Westover.
Detecting protein sequence conservation via metric embeddings.
Bioinformatics, 19(Suppl. 1):i122–i129, 2003"""
xx = parasail.nw_stats(s1, s1, open=3, extend=3, matrix=parasail.blosum62).score
yy = parasail.nw_stats(s2, s2, open=3, extend=3, matrix=parasail.blosum62).score
xy = parasail.nw_stats(s1, s2, open=3, extend=3, matrix=parasail.blosum62).score
D = xx + yy - 2 * xy
return D
def nw_metric(self, s1, s2):
"""May or may not produce a true metric. Details in:
E. Halpering, J. Buhler, R. Karp, R. Krauthgamer, and B. Westover.
Detecting protein sequence conservation via metric embeddings.
Bioinformatics, 19(Suppl. 1):i122–i129, 2003"""
xx = parasail.nw_stats(s1, s1, open=3, extend=3,
matrix=parasail.blosum62).score
yy = parasail.nw_stats(s2, s2, open=3, extend=3,
matrix=parasail.blosum62).score
xy = parasail.nw_stats(s1, s2, open=3, extend=3,
matrix=parasail.blosum62).score
D = xx + yy - 2 * xy
return D
def nw_hamming_metric(s1, s2, matrix='blosum62', open=3, extend=3):
"""Function applying Parasail's Needleman-Wuncsh Algorithm to align and
compute a Hamming Distance between any two sequences: number of
mismatched positions. Gaps count as a mismatch. Penalties and matrix
are used for alignment purposes, not in the distance calculation.
Parameters
----------
s1: string
string containing amino acid letters
s2: string
string containing amino acid letters
matrix : str
Attribute of parasail that names a substitution matrix
Returns
-------
D : float
distance between strings (Hamming Distance: number of mismatched positions)
Notes
-----
.. code-block:: python
xy = parasail.nw_stats(s1, s2, open=open, extend=extend, matrix=matrix)
xy_t = parasail.nw_trace(s1, s2, open=open, extend=extend, matrix=matrix)
hamming_distance = len(xy_t.traceback.comp)-xy.matches
return hamming_distance"""
p_matrix = getattr(parasail, matrix)
xy = parasail.nw_stats(s1, s2, open=open, extend=extend, matrix=p_matrix)
xy_t = parasail.nw_trace(s1, s2, open=open, extend=extend, matrix=p_matrix)
D = len(xy_t.traceback.comp)-xy.matches
return D
def _try_cache(self, e):
try:
xx = self.sim_cache[(e, e)]
except KeyError:
xx = parasail.nw_stats(e, e, **self.paraParams).score
self.sim_cache[(e, e)] = xx
return xx
def _try_cache(self, e):
try:
xx = self.sim_cache[(e, e)]
except KeyError:
xx = parasail.nw_stats(e, e, **self.paraParams).score
self.sim_cache[(e, e)] = xx
return xx
def nw_metric(s1, s2, matrix='blosum62', open=3, extend=3, return_similarity=False):
"""Function applying Parasail's Needleman-Wuncsh Algorithm to compute
a distance between any two sequences.
Parameters
----------
s1: string
string containing amino acid letters
s2: string
string containing amino acid letters
matrix : str
Attribute of parasail that names a substitution matrix
Returns
-------
D : float
distance via reciprocal alignment scores.
Notes
-----
.. code-block:: python
xx = parasail.nw_stats(s1, s1, open=open, extend=extend, matrix=matrix).score
yy = parasail.nw_stats(s2, s2, open=open, extend=extend, matrix=matrix).score
xy = parasail.nw_stats(s1, s2, open=open, extend=extend, matrix=matrix).score
D = xx + yy - 2 * xy
return D
May or may not produce a true metric. Details in:
E. Halpering, J. Buhler, R. Karp, R. Krauthgamer, and B. Westover.
Detecting protein sequence conservation via metric embeddings.
Bioinformatics, 19 (sup 1) 2003
"""
p_matrix = getattr(parasail, matrix)
xy = parasail.nw_stats(s1, s2, open=open, extend=extend, matrix=p_matrix).score
if return_similarity:
return xy
else:
xx = parasail.nw_stats(s1, s1, open=open, extend=extend, matrix=p_matrix).score
yy = parasail.nw_stats(s2, s2, open=open, extend=extend, matrix=p_matrix).score
D = xx + yy - 2 * xy
return D
def calculate(self,
query_base_sequence,
target_base_sequence,
gap_open=16,
gap_extend=4):
result = parasail.nw_stats(query_base_sequence, target_base_sequence,
gap_open, gap_extend, parasail.dnafull)
score = result.similar * 1.0 / result.length
return score
def metric(self, i1, i2):
"""sklearn specifies that function will receive
two rows as parameters and return one value as distance"""
xx = self._try_cache(self.i2e[i1[0]])
yy = self._try_cache(self.i2e[i2[0]])
"""Don't need to cache the xy similarity because it doesn't have other uses"""
xy = parasail.nw_stats(self.i2e[i1[0]], self.i2e[i2[0]],
**self.paraParams).score
D = xx + yy - 2 * xy
return D
def metric(self, i1, i2):
"""sklearn specifies that function will receive
two rows as parameters and return one value as distance"""
xx = self._try_cache(self.i2e[i1[0]])
yy = self._try_cache(self.i2e[i2[0]])
"""Don't need to cache the xy similarity because it doesn't have other uses"""
xy = parasail.nw_stats(self.i2e[i1[0]], self.i2e[i2[0]], **self.paraParams).score
D = xx + yy - 2 * xy
return D
def hm_metric(s1, s2, matrix='blosum62', open=3, extend=3):
"""
Function applying Parasail's Needleman-Wuncsh Algorithm to allign and get
a Hamming Distance between any two sequences: number of mismatched positions
Parameters
----------
s1: string
string containing amino acid letters
s2: string
string containing amino acid letters
Returns
-------
D : float
distance between strings (Hamming Distance: number of mismatched positions)
Notes
-----
.. code-block:: python
xy = parasail.nw_stats(s1, s2, open=open, extend=extend, matrix=matrix)
xy_t = parasail.nw_trace(s1, s2, open=open, extend=extend, matrix=matrix)
hamming_distance = len(xy_t.traceback.comp)-xy.matches
return hamming_distance
"""
p_matrix = getattr(parasail, matrix)
xy = parasail.nw_stats(s1, s2, open=open, extend=extend, matrix=p_matrix)
xy_t = parasail.nw_trace(s1, s2, open=open, extend=extend, matrix=p_matrix)
D = len(xy_t.traceback.comp) - xy.matches
return D
def hm_matches(s1, s2, matrix=parasail.blosum62, open=3, extend=3):
xy = parasail.nw_stats(s1, s2, open=open, extend=extend, matrix=matrix)
xy_t = parasail.nw_trace(s1, s2, open=open, extend=extend, matrix=matrix)
return xy.matches
def aligned_mm_metric(s1, s2, open=3, extend=3, matrix=parasail.blosum62):
res = parasail.nw_stats(s1, s2, open=open, extend=extend, matrix=matrix)
return res.length - res.matches
