def summary_allele(qbase):
"Counts the number of reads for each allele in genotype. Return the two most expressed"
base_count = {
'A': qbase.count('A'),
'C': qbase.count('C'),
'G': qbase.count('G'),
'T': qbase.count('T'),
'N': qbase.count('N')
}
call = heapq.nlargest(3, base_count.iteritems(), heapq.itemgetter(1))
return (call)
def getResults(self, top_n = -1):
"""Returns the n kmers which occur most frequently,
input: a kmer dictionary with occurrence as values,
returns: a set of pairs with the n most frequent kmers"""
if self.results is None:
self.findMotifs()
if top_n == -1:
top_n = len(self.results)
kmers = {}
for i in xrange(len(self.results)):
kmers[self.results[i].getMotif()] = self.results[i].getScore()
return nlargest(top_n, kmers.iteritems(), itemgetter(1))
def getResults(self, top_n=-1):
"""Returns the n kmers which occur most frequently,
input: a kmer dictionary with occurrence as values,
returns: a set of pairs with the n most frequent kmers"""
if self.results is None:
self.findMotifs()
if top_n == -1:
top_n = len(self.results)
kmers = {}
for i in xrange(len(self.results)):
kmers[self.results[i].getMotif()] = self.results[i].getScore()
return nlargest(top_n, kmers.iteritems(), itemgetter(1))
def countKmers(self, top_n = -1):
"""Puts all k-mers occurring in a string into a dictionary
and returns it, with values containing the number of occurrences."""
kmers = {}
for seq in self.seqs.getAllSequences():
seq = seq.upper()
for i in xrange(len(seq)):
kmer = seq[i:i + self.getKmerLength()]
if len(kmer) == self.getKmerLength():
if kmer in kmers:
kmers[kmer] += 1
else:
kmers[kmer] = 1
if top_n == -1:
return kmers
else:
top_list = nlargest(top_n, kmers.iteritems(), itemgetter(1))
top_dict = {}
for key, value in top_list:
top_dict[key] = value
return top_dict
def countKmers(self, top_n=-1):
"""Puts all k-mers occurring in a string into a dictionary
and returns it, with values containing the number of occurrences."""
kmers = {}
for seq in self.seqs.getAllSequences():
seq = seq.upper()
for i in xrange(len(seq)):
kmer = seq[i:i + self.getKmerLength()]
if len(kmer) == self.getKmerLength():
if kmer in kmers:
kmers[kmer] += 1
else:
kmers[kmer] = 1
if top_n == -1:
return kmers
else:
top_list = nlargest(top_n, kmers.iteritems(), itemgetter(1))
top_dict = {}
for key, value in top_list:
top_dict[key] = value
return top_dict
def summary_allele(qbase):
"Counts the number of reads for each allele in genotype. Return the two most expressed"
base_count= {'A':qbase.count('A'), 'C':qbase.count('C'), 'G':qbase.count('G'), 'T':qbase.count('T'), 'N':qbase.count('N')}
call= heapq.nlargest(3, base_count.iteritems(), heapq.itemgetter(1))
return(call)
