1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
|
import cython
import numpy as np
cimport numpy as np
@cython.boundscheck(False)
@cython.wraparound(False)
def smith_waterman(unicode string1, unicode string2, float gap_cost, \
sim_func):
cdef int i = 0, j = 0
cdef double match = 0.0, delete = 0.0, insert = 0.0
cdef double sim_score = 0.0, max_value = 0.0
cdef int len_s1 = len(string1), len_s2 = len(string2)
cdef double[:,:] dist_mat = np.zeros((len(string1) + 1, len(string2) + 1), dtype=float)
# Smith Waterman DP calculations
for i from 1 <= i < (len_s1 + 1):
for j from 1 <= j < (len_s2 + 1):
sim_func_score = sim_func(string1[i - 1], string2[j - 1])
match = dist_mat[i - 1, j - 1] + sim_func_score
delete = dist_mat[i - 1, j] - gap_cost
insert = dist_mat[i, j - 1] - gap_cost
dist_mat[i, j] = max(0, match, delete, insert)
max_value = max(max_value, dist_mat[i, j])
return max_value
|
