def linear_seqs_division(seqs, ge , s ):
"""
seqs is array of seq that we want to match
ge - gap extextion penalyty
s is substitution matrix or array [x,y] x is cost of matching
when symbols that we are matching are the same and y if they are not
function return array that contain divion of seq in seqs into
two smaller string second string is reverse
"""
H = init_matrix_H([len(seq) + 1 for seq in seqs], ge)
d = create_matching_function(s , ge)
P = fill_half(H ,seqs, ge, d)
a = len(seqs[0])//2
B = np.argwhere(P == np.max(P))
array = []
for b in B:
b = b.item()
seqs_0_begin = seqs[0][0:a]
seqs_1_begin = seqs[1][0:b]
seqs_0_end = seqs[0][a:]
seqs_1_end = seqs[1][b:]
array.append(([seqs_0_begin,seqs_1_begin],[seqs_0_end[::-1],seqs_1_end[::-1]]))
return array
def multidimesional_N_W_algoritm(seqs, g, s, max_mathing):
"""
seqs is array of seq that we want to match
g is gap extention penalty
s is substitution matrix or array [x,y] x is cost of matching
when symbols that we are matching are the same and y if they are not
max_mathing is maximal number of matching that are returned
function return pair containg cost of matching and matches
"""
H = init_matrix([len(seq) + 1 for seq in seqs], g)
d = create_matching_function(s, g)
fill_H_all(H, seqs, d)
return H[tuple([len(seq)
for seq in seqs])], matching(H, seqs, d, max_mathing)
def linear_gap_algorytm(seqs, go, ge, s, max_number_of_matching):
"""
seqs is array of seq that we want to match
g is gap extention cost
s is substitution matrix or array [x,y] x is cost of matching
when symbols that we are matching are the same and y if they are not
max_mathing is maximal number of matching that are returned
function return pair containg cost of matching and matches
"""
H = init_matrix_H([len(seq) + 1 for seq in seqs], go, ge)
E = init_matrix_E([len(seq) + 1 for seq in seqs], go, ge)
F = init_matrix_F([len(seq) + 1 for seq in seqs], go, ge)
d = create_matching_function(s, ge)
fill_all(H, E, F, seqs, go, ge, d)
X = matching(H, E, F, seqs, d, go, ge, max_number_of_matching)
#X = list(np.unique(X))
return (H[(len(seqs[0]), len(seqs[1]))], X)
"""
#print((len(seqs[0]) + 1) * (len(seqs [1]) + 1))
#print("elo")
if((len(seqs[0]) + 1) * (len(seqs [1]) + 1) <= L or len(seqs[0]) < 10):
return multidimesional_N_W_algoritm(seqs, ge , s, max_number_of_matching)
else:
array = linear_seqs_division(seqs, ge , s)
maximum = -float("inf")
for seqs1, seqs_rev1 in array:
score_begin, matches_begin = linear_algorytm(seqs1, ge , s, max_number_of_matching, L)
score_end, matches_end = linear_algorytm(seqs_rev1, ge , s, max_number_of_matching, L)
if (maximum < score_begin +score_end):
maximum = score_begin +score_end
max_matches_begin = matches_begin
max_matches_end = matches_end
return maximum, merge(max_matches_begin, max_matches_end, max_number_of_matching)
if __name__ == "__main__":
s1= "AAABB"
s2= "AAA"
seqs = [s1, s2 ]
ge = - 1
d = create_matching_function([1,-1] , ge)
H = init_matrix_H([len(seq) + 1 for seq in seqs] ,ge)
X = linear_algorytm(seqs, ge , [1,-1], 100, 7*2)
score = X[0]
if(not score == 1):
raise AssertionError