desc[ens]=lst[7]
ucsc,tkname=sys.argv[1:]
# dump
fout=open(tkname,'w')
fout2=open(tkname+'_load','w')
id=1
with open(ucsc) as fin:
for line in fin:
lst=line.rstrip().split('\t')
g=parseUcscgenestruct.parse(lst,True)
name=lst[1]
fout.write('{0}\t{1}\t{2}\tname:"{3}",id:{4},strand:"{5}",'.format(
g['chrom'],
g['start'],
g['stop'],
name,
id,
g['strand']))
id+=1
if 'thin' in g or 'thick' in g:
fout.write('struct:{')
if 'thin' in g:
fout.write('thin:[')
for x in g['thin']:
fout.write('[{0},{1}],'.format(x[0],x[1]))
desc[lst[2]] = w
desc[lst[3]] = w
symbol[lst[2]] = lst[0]
symbol[lst[3]] = lst[0]
i += 1
print 'refLink: ' + str(i)
# dump
fout = open(tkname, 'w')
fout2 = open(tkname + '_load', 'w')
id = 1
with open(ucsc) as fin:
for line in fin:
lst = line.rstrip().split('\t')
g = parseUcscgenestruct.parse(lst, True)
name = lst[1]
fout.write('{0}\t{1}\t{2}\tname:"{3}",id:{4},strand:"{5}",'.format(
g['chrom'], g['start'], g['stop'], name, id, g['strand']))
id += 1
if 'thin' in g or 'thick' in g:
fout.write('struct:{')
if 'thin' in g:
fout.write('thin:[')
for x in g['thin']:
fout.write('[{0},{1}],'.format(x[0], x[1]))
fout.write('],')
if 'thick' in g:
fout.write('thick:[')
for x in g['thick']:
fout.write('[{0},{1}],'.format(x[0], x[1]))
# dump
fout=open(tkname,'w')
fout2=open(tkname+'_load','w')
id=1
with open('wgEncodeGencodeAttrsV17.txt') as fin:
for line in fin:
lst=line.rstrip().split('\t')
transcript=lst[4]
if transcript not in genes: continue
g=parseUcscgenestruct.parse(genes[transcript],True)
fout.write('{0}\t{1}\t{2}\tname:"{3}",id:{4},strand:"{5}",'.format(
g['chrom'],
g['start'],
g['stop'],
lst[1] if len(lst[1])>0 else transcript,
id,
g['strand']))
id+=1
if 'thin' in g or 'thick' in g:
fout.write('struct:{')
if 'thin' in g:
fout.write('thin:[')
for x in g['thin']:
fout.write('[{0},{1}],'.format(x[0],x[1]))
fout.write('],')
