def multithreadedJaccardBed(*bedSortedPairs):
# bedSortedPairs = jaccardBed()
jaccardScores = dict()
errorBedSortedPairs = []
try:
if (os.stat(bedSortedPairs[0][0]).st_size >= os.stat(
bedSortedPairs[0][1]).st_size):
print('true')
file1 = BedTool(bedSortedPairs[0][0])
file2 = BedTool(bedSortedPairs[0][1])
jaccardResults = BedTool.jaccard(file1, file2)
jaccardScore = jaccardResults['jaccard']
# jaccardScore = 0
scoredPair = (file1, file2)
jaccardScores[str(scoredPair)] = jaccardScore
else:
print('yuck')
file1 = BedTool(bedSortedPairs[0][1])
file2 = BedTool(bedSortedPairs[0][0])
jaccardResults = BedTool.jaccard(file1, file2)
jaccardScore = jaccardResults['jaccard']
# jaccardScore = 0
scoredPair = (file1, file2)
jaccardScores[str(scoredPair)] = jaccardScore
'''
jaccardResults = BedTool.jaccard(file1, file2)
jaccardScore = jaccardResults['jaccard']
scoredPair = (file1, file2)
jaccardScores[str(scoredPair)] = jaccardScore
'''
except Exception as error:
if (os.stat(bedSortedPairs[0][0]).st_size >= os.stat(
bedSortedPairs[0][1]).st_size):
print('ok!')
print('***')
print(bedSortedPairs[0][0])
print(bedSortedPairs[0][1])
print('***')
else:
print('youch!')
print('-----')
print(bedSortedPairs[0][0])
print(bedSortedPairs[0][1])
print('-----')
errorBedSortedPair = [file1, file2]
errorBedSortedPairs.append(errorBedSortedPair)
print('error')
print(error)
return jaccardScores, errorBedSortedPairs
def run_jaccard(fileA, fileB, genomefile):
""" Running bedtools. Reads in two bedtools approved file types, sorts the files,
and calculates a jaccard score.
"""
a = BedTool(fileA)
a = a.sort(g=genomefile)
b = BedTool(fileB)
b = b.sort(g=genomefile)
j = a.jaccard(b, g=genomefile)
j["fileA"] = fileA.split("/")[-1]
j["fileB"] = fileB.split("/")[-1]
keylist = list(j.keys())
keylist.sort()
data = [str(j[key]) for key in keylist]
return (data, keylist)
def multithreadedJaccardBed(*bedSortedPairs):
# bedSortedPairs = jaccardBed()
jaccardScores = dict()
errorBedSortedPairs = []
try:
file1 = BedTool(bedSortedPairs[0][0])
file2 = BedTool(bedSortedPairs[0][1])
jaccardResults = BedTool.jaccard(file1, file2)
jaccardScore = jaccardResults['jaccard']
scoredPair = (file1, file2)
jaccardScores[str(scoredPair)] = jaccardScore
except Exception as error:
errorBedSortedPair = [file1, file2]
errorBedSortedPairs.append(errorBedSortedPair)
# print('error')
# print(error)
return jaccardScores, errorBedSortedPairs
def sequentialJaccardBed():
sequential_time = time.time()
bedSortedPairs = jaccardBed()
jaccardScores = []
errorBedSortedPairs = []
for pair in bedSortedPairs:
try:
file1 = BedTool(pair[0])
file2 = BedTool(pair[1])
jaccardResults = BedTool.jaccard(file1, file2)
jaccardScores.append(jaccardResults['jaccard'])
except Exception as error:
errorBedSortedPair = [file1, file2]
errorBedSortedPairs.append(errorBedSortedPair)
# print('error')
# print(error)
print('Sequentially calculating jaccard scores --- %.2f seconds ---' %
(time.time() - sequential_time))
cons_bedgraph = BedTool('/vol1/opt/data/hg19.100way.phyloP100way.bg.gz')
header_fields = ['#cell.type.1','cell.type.2','type','jaccard','mean.cons']
print '\t'.join(header_fields)
for fname1, fname2 in combinations(subset, r=2):
ctype1 = get_cell_type(fname1)
ctype2 = get_cell_type(fname2)
tool1 = BedTool(fname1)
tool2 = BedTool(fname2)
# jaccard statistic
result = tool1.jaccard(tool2, f=0.5, r=False)
stat = result['jaccard']
# conservation measurement
result1 = tool1.intersect(tool2, v=True)
result2 = tool2.intersect(tool1, v=True)
cons1 = result1.map(cons_bedgraph, o='mean', c=4)
cons2 = result2.map(cons_bedgraph, o='mean', c=4)
pbd.set_trace()
fields = [ctype1, ctype2, stat]
print '\t'.join(map(str, fields))
subset = random.sample(filenames, 10)
cons_bedgraph = BedTool('/vol1/opt/data/hg19.100way.phyloP100way.bg.gz')
header_fields = ['#cell.type.1', 'cell.type.2', 'type', 'jaccard', 'mean.cons']
print '\t'.join(header_fields)
for fname1, fname2 in combinations(subset, r=2):
ctype1 = get_cell_type(fname1)
ctype2 = get_cell_type(fname2)
tool1 = BedTool(fname1)
tool2 = BedTool(fname2)
# jaccard statistic
result = tool1.jaccard(tool2, f=0.5, r=False)
stat = result['jaccard']
# conservation measurement
result1 = tool1.intersect(tool2, v=True)
result2 = tool2.intersect(tool1, v=True)
cons1 = result1.map(cons_bedgraph, o='mean', c=4)
cons2 = result2.map(cons_bedgraph, o='mean', c=4)
pbd.set_trace()
fields = [ctype1, ctype2, stat]
print '\t'.join(map(str, fields))
def score(onlyfiles): for bed in onlyfiles: a = BedTool(mypath+bed) jac=BedTool.jaccard(userfile.sort(),a) if jac['jaccard']>jaccard: scores[bed]=jac['jaccard']
