def writesets(datasets,outputdir,outformat='std'):
"""write data with all the combinations into file """
if outformat=='std':
namlis=[ele[0] for ele in datasets]
datalis=[list(ele[1]) for ele in datasets]
filename=os.path.join(outputdir,'stdformat_'+'_'.join(namlis)+'.txt')
datatrsp=fhutils.transpose(datalis,'NA')
out=open(filename,'w')
out.write('\t'.join(namlis)+'\n')
for row in datatrsp:
out.write('\t'.join([str(x) for x in row])+'\n')
out.close()
return filename
elif outformat=='evenn':
tabnames=[]
for name,data in datasets:
out=open(os.path.join(outputdir,name+'_table.txt'),'w')
tabnames.append(name+'_table.txt')
data=list(set(data))
data.insert(0,name)
for row in data:
if isinstance(row,list):
out.write('\t'.join([str(x) for x in row])+'\n')
else:
out.write(str(row)+'\n')
out.close()
return tabnames
elif outformat=='venneuler':
totalitems=[]
hdnames=[]
for name,data in datasets:
totalitems.append(dict([(ele,name) for ele in data]))
hdnames.append(name)
totalids=fhutils.flatten([list(d.keys()) for d in totalitems])
tablecontent=[]
for k in set(totalids):
outrows=[k]
for d in totalitems:
if k in d:
outrows.append('1')
else:
outrows.append('0')
tablecontent.append(outrows)
out=open(os.path.join(outputdir,name+'_table.txt'),'w')
out.write('id\t'+'\t'.join(hdnames)+'\n')
for row in tablecontent:
out.write('\t'.join([str(x) for x in row])+'\n')
out.close()
veninputfile = os.path.join(outputdir,name+'_table.txt')
return veninputfile
else:
print 'nothing to do'
def writesets(datasets, outputdir, outformat='std'):
"""write data with all the combinations into file """
if outformat == 'std':
namlis = [ele[0] for ele in datasets]
datalis = [list(ele[1]) for ele in datasets]
filename = os.path.join(outputdir,
'stdformat_' + '_'.join(namlis) + '.txt')
datatrsp = fhutils.transpose(datalis, 'NA')
out = open(filename, 'w')
out.write('\t'.join(namlis) + '\n')
for row in datatrsp:
out.write('\t'.join([str(x) for x in row]) + '\n')
out.close()
return filename
elif outformat == 'evenn':
tabnames = []
for name, data in datasets:
out = open(os.path.join(outputdir, name + '_table.txt'), 'w')
tabnames.append(name + '_table.txt')
data = list(set(data))
data.insert(0, name)
for row in data:
if isinstance(row, list):
out.write('\t'.join([str(x) for x in row]) + '\n')
else:
out.write(str(row) + '\n')
out.close()
return tabnames
elif outformat == 'venneuler':
totalitems = []
hdnames = []
for name, data in datasets:
totalitems.append(dict([(ele, name) for ele in data]))
hdnames.append(name)
totalids = fhutils.flatten([list(d.keys()) for d in totalitems])
tablecontent = []
for k in set(totalids):
outrows = [k]
for d in totalitems:
if k in d:
outrows.append('1')
else:
outrows.append('0')
tablecontent.append(outrows)
out = open(os.path.join(outputdir, name + '_table.txt'), 'w')
out.write('id\t' + '\t'.join(hdnames) + '\n')
for row in tablecontent:
out.write('\t'.join([str(x) for x in row]) + '\n')
out.close()
veninputfile = os.path.join(outputdir, name + '_table.txt')
return veninputfile
else:
print 'nothing to do'
def venn_runner(dataset,outputfilename='vennrunner.txt',vennengine='',outputdir=''):
outputdir = os.path.abspath(outputdir)
outputfilename=os.path.join(outputdir,outputfilename)
logger.info('#########\nvenn run started#########\n')
if isinstance(dataset,list):
datadic=dict(dataset)
uniqueinfo=[]
commoninfo=[]
for i in range(0,len(datadic)):
for v in itertools.combinations(datadic.keys(),i+1):
combo=[set(datadic[x]) for x in v ]
tmp=combo[0]
excl=combo[0]
for ele in combo[1:]:
tmp=tmp&ele
excl=excl-ele
commoninfo.append((v,tmp))
uniqueinfo.append((v,excl))
tabinfo=[]
for name,ele in commoninfo:
logger.info('total in {name}: {lenset}'.format(name=' & '.join(name),lenset=len(ele)))
tabinfo.append(['_&_'.join(name)]+list(ele))
for name,ele in uniqueinfo:
logger.info('total in {name}: {lenset}'.format(name=' not '.join(name),lenset=len(ele)))
tabinfo.append([' not '.join(name)]+list(ele))
tabinfotrsp=fhutils.transpose(tabinfo,'')
out=open(outputfilename,'w')
for ele in tabinfotrsp:
out.write('\t'.join([str(x) for x in ele])+'\n')
out.close()
if vennengine=='evenn':
logger.error('NOT WORKING EVENN')
pyevenn(dataset,figurename=fhutils.renamefilename(outputfilename,midtag='_'+vennengine,suffix='png'),outputdir =outputdir)
elif vennengine=='rvenn':
pyrvenn(dataset,figurename=fhutils.renamefilename(outputfilename,midtag='_'+vennengine,suffix='tiff'),outputdir = outputdir)
elif vennengine=='venneuler':
pyvenneuler(dataset,figurename=fhutils.renamefilename(outputfilename,midtag='_'+vennengine,suffix='pdf'),outputdir=outputdir)
elif vennengine=='rvennerable':
logger.error('NOT WORKING rvennerable')
pyrvennerable(dataset,figurename=fhutils.renamefilename(outputfilename,midtag='_'+vennengine,suffix='pdf'),outputdir=outputdir)
else:
print 'no venn pdf'
def venn_runner(dataset,
outputfilename='vennrunner.txt',
vennengine='',
outputdir=''):
outputdir = os.path.abspath(outputdir)
outputfilename = os.path.join(outputdir, outputfilename)
logger.info('#########\nvenn run started#########\n')
if isinstance(dataset, list):
datadic = dict(dataset)
uniqueinfo = []
commoninfo = []
for i in range(0, len(datadic)):
for v in itertools.combinations(datadic.keys(), i + 1):
combo = [set(datadic[x]) for x in v]
tmp = combo[0]
excl = combo[0]
for ele in combo[1:]:
tmp = tmp & ele
excl = excl - ele
commoninfo.append((v, tmp))
uniqueinfo.append((v, excl))
tabinfo = []
for name, ele in commoninfo:
logger.info('total in {name}: {lenset}'.format(name=' & '.join(name),
lenset=len(ele)))
tabinfo.append(['_&_'.join(name)] + list(ele))
for name, ele in uniqueinfo:
logger.info('total in {name}: {lenset}'.format(name=' not '.join(name),
lenset=len(ele)))
tabinfo.append([' not '.join(name)] + list(ele))
tabinfotrsp = fhutils.transpose(tabinfo, '')
out = open(outputfilename, 'w')
for ele in tabinfotrsp:
out.write('\t'.join([str(x) for x in ele]) + '\n')
out.close()
if vennengine == 'evenn':
logger.error('NOT WORKING EVENN')
pyevenn(dataset,
figurename=fhutils.renamefilename(outputfilename,
midtag='_' + vennengine,
suffix='png'),
outputdir=outputdir)
elif vennengine == 'rvenn':
pyrvenn(dataset,
figurename=fhutils.renamefilename(outputfilename,
midtag='_' + vennengine,
suffix='tiff'),
outputdir=outputdir)
elif vennengine == 'venneuler':
pyvenneuler(dataset,
figurename=fhutils.renamefilename(outputfilename,
midtag='_' + vennengine,
suffix='pdf'),
outputdir=outputdir)
elif vennengine == 'rvennerable':
logger.error('NOT WORKING rvennerable')
pyrvennerable(dataset,
figurename=fhutils.renamefilename(outputfilename,
midtag='_' +
vennengine,
suffix='pdf'),
outputdir=outputdir)
else:
print 'no venn pdf'