def string_spelled_by_gapped_patterns(gapped_patterns, k, d):
first_patterns = [gp[0] for gp in gapped_patterns]
second_patterns = [gp[1] for gp in gapped_patterns]
prefix_string = kmer_reconstruction.reconstruct(first_patterns)
suffix_string = kmer_reconstruction.reconstruct(second_patterns)
for i in range(k + d + 1, len(prefix_string)):
p = prefix_string[i]
s = suffix_string[i - k - d]
if p != s:
return None
return prefix_string + suffix_string[-(k + d):]
def create_circular_string(k):
format_string = '{0:0' + str(k) + 'b}'
strings = [format_string.format(x) for x in range(0, 2 ** k)]
graph = debruijn.create_from_patterns(strings)
cycle = eulerian.cycle.find_cycle(graph)[:-(k-1)]
text = kmer_reconstruction.reconstruct(cycle)
return text
def reconstruct(kmers):
graph = debruijn.create_from_patterns(kmers)
path = eulerian.path.find_path(graph)
text = kmer_reconstruction.reconstruct(path)
return text
import contigs
import kmer_reconstruction
import sys
import debruijn
filename = sys.argv[1]
with open(filename, 'r') as f:
lines = [l.strip() for l in f.readlines() if l.strip() != '']
graph = debruijn.create_from_patterns(lines)
res = contigs.maximal_non_branching_paths(graph)
cgs = [kmer_reconstruction.reconstruct(ks) for ks in res]
output = '\n'.join(cgs)
print(output)
import kmer_reconstruction, sys
filename = sys.argv[1]
with open(filename, 'r') as f:
kmers = [line.strip() for line in f.readlines() if line.strip() != '']
print(kmer_reconstruction.reconstruct(kmers))
def test_single_kmer():
kmers = ['ACCGA']
expected = 'ACCGA'
assert (expected == reconstruct(kmers))
def test_normal_case():
kmers = ['ACCGA', 'CCGAA', 'CGAAG', 'GAAGC', 'AAGCT']
expected = 'ACCGAAGCT'
assert (expected == reconstruct(kmers))
def test_empty_kmers():
kmers = []
expected = ''
assert (expected == reconstruct(kmers))
def test_single__char_kmers():
kmers = ['A', 'C', 'C', 'G', 'A']
expected = 'ACCGA'
assert (expected == reconstruct(kmers))