def read_sorted_bedgraphs(directory, transcript_dict):
stranded_bedgraphs = {}
for file in os.listdir(directory):
if file.endswith("_CNAGsort.bedgraph"):
if "plus" in file:
if file.split('_plus')[0] not in stranded_bedgraphs:
stranded_bedgraphs[file.split('_plus')[0]] = [None, None]
stranded_bedgraphs[file.split('_plus')[0]][0] = SP.read_CNAGsort_bedgraph2(file, transcript_dict, organism='pombe')
elif 'minus' in file:
if file.split('_minus')[0] not in stranded_bedgraphs:
stranded_bedgraphs[file.split('_minus')[0]] = [None, None]
stranded_bedgraphs[file.split('_minus')[0]][1] = SP.read_CNAGsort_bedgraph2(file, transcript_dict, organism='pombe')
return stranded_bedgraphs
def find_polyA_sites(transcript_dict, window=220):
polyA_bg = SP.read_CNAGsort_bedgraph2(
'/home/jordan/GENOMES/POMBE/polyA_sites_CNAGsort.bedgraph',
transcript_dict,
organism='pombe')
pA_dict = {}
for tx, s in polyA_bg.iteritems():
s = s[s > 0]
if len(s) > 0:
if transcript_dict[tx][2] == '+':
#pA_site = max(s.index)
s.sort_values(ascending=False, inplace=True)
pA_site = s.index[0]
pA_dict[tx] = [
pA_site - window, pA_site + window, transcript_dict[tx][2],
transcript_dict[tx][3]
]
elif transcript_dict[tx][2] == '-':
#pA_site = min(s.index)
s.sort_values(ascending=False, inplace=True)
pA_site = s.index[0]
pA_dict[tx] = [
pA_site - window, pA_site + window, transcript_dict[tx][2],
transcript_dict[tx][3]
]
return pA_dict
def read_sorted_bedgraphs(directory, transcript_dict):
stranded_bedgraphs = {}
for file in os.listdir(directory):
if file.endswith("_CNAGsort.bedgraph"):
if "plus" in file:
if file.split('_plus')[0] not in stranded_bedgraphs:
stranded_bedgraphs[file.split('_plus')[0]] = [None, None]
stranded_bedgraphs[file.split('_plus')
[0]][0] = SP.read_CNAGsort_bedgraph2(
file, transcript_dict, organism='pombe')
elif 'minus' in file:
if file.split('_minus')[0] not in stranded_bedgraphs:
stranded_bedgraphs[file.split('_minus')[0]] = [None, None]
stranded_bedgraphs[file.split('_minus')
[0]][1] = SP.read_CNAGsort_bedgraph2(
file, transcript_dict, organism='pombe')
return stranded_bedgraphs
def find_polyA_sites(transcript_dict, window=220):
polyA_bg = SP.read_CNAGsort_bedgraph2('/home/jordan/GENOMES/POMBE/polyA_sites_CNAGsort.bedgraph', transcript_dict, organism='pombe')
pA_dict = {}
for tx, s in polyA_bg.iteritems():
s = s[s > 0]
if len(s) > 0:
if transcript_dict[tx][2] == '+':
#pA_site = max(s.index)
s.sort_values(ascending=False, inplace=True)
pA_site = s.index[0]
pA_dict[tx] = [pA_site-window, pA_site+window, transcript_dict[tx][2], transcript_dict[tx][3]]
elif transcript_dict[tx][2] == '-':
#pA_site = min(s.index)
s.sort_values(ascending=False, inplace=True)
pA_site = s.index[0]
pA_dict[tx] = [pA_site-window, pA_site+window, transcript_dict[tx][2], transcript_dict[tx][3]]
return pA_dict
