68 lines
1.6 KiB
Python
68 lines
1.6 KiB
Python
"""
|
|
Small assembly module based on de bruijn graphs
|
|
"""
|
|
import networkx as nx
|
|
|
|
def read_fastq(fichier):
|
|
"""
|
|
Arguments:
|
|
fichier, str: path to fastq file
|
|
|
|
Returns:
|
|
a str generator, generator of sequences
|
|
"""
|
|
with open(fichier, 'r') as filin:
|
|
for line in filin:
|
|
yield filin.readline().strip()
|
|
filin.readline()
|
|
filin.readline()
|
|
|
|
def cut_kmer(seq, k):
|
|
"""
|
|
Arguments:
|
|
seq, str: a sequence
|
|
k, int: k-mer size, must be shorter than len(seq)
|
|
|
|
Returns:
|
|
an iterator returning str
|
|
"""
|
|
for i in range(len(seq)-(k-1)):
|
|
yield seq[i:i+k]
|
|
|
|
def build_kmer_dict(fichier, k):
|
|
"""
|
|
Arguments:
|
|
fichier, str: path to fastq file
|
|
k, int: k-mer size, must be shorter than len(seq)
|
|
|
|
Return:
|
|
hash_table, dict: dictionnary with key = k-mer as str
|
|
and value count of k-mer occurence
|
|
"""
|
|
hash_table = {}
|
|
it_fastq = read_fastq(fichier)
|
|
for seq in it_fastq:
|
|
it_kmer = cut_kmer(seq, k)
|
|
for kmer in it_kmer:
|
|
try:
|
|
hash_table[kmer]
|
|
except KeyError:
|
|
hash_table[kmer] = 1
|
|
else:
|
|
hash_table[kmer] += 1
|
|
|
|
return hash_table
|
|
|
|
def build_graph(hash_table):
|
|
"""
|
|
Arguments:
|
|
hash_table, dict: dictionnary obtained with build_kmer_dict() function
|
|
Return:
|
|
graph, nx.Graph: the de Bruijn tree corresponding to hash_table
|
|
"""
|
|
graph = nx.DiGraph()
|
|
for key in hash_table:
|
|
graph.add_edge(key[:-1], key[1:], weight=hash_table[key])
|
|
|
|
return graph
|