def assemble(options):
# check output directory
out_dir = check_dir(options['<circ_dir>'])
# check tophat results
if options['--tophat-dir']:
tophat_dir = check_dir(options['--tophat-dir'])
else:
tophat_dir = check_dir(out_dir + '/tophat')
# check cufflinks
if which('cufflinks') is None:
sys.exit('Cufflinks is required for CIRCexplorer2 assemble!')
# check genePredToGtf
if which('genePredToGtf') is None:
sys.exit('genePredToGtf is required for CIRCexplorer2 assemble!')
# check gtfToGenePred
if which('gtfToGenePred') is None:
sys.exit('gtfToGenePred is required for CIRCexplorer2 assemble!')
# prepare cufflinks directory
cufflinks_dir = out_dir + '/cufflinks'
create_dir(cufflinks_dir)
# filter ref file
ref_filter(options['--ref'], tophat_dir, out_dir)
# assemble with cufflinks
cufflinks_assemble(out_dir, tophat_dir, cufflinks_dir, options['--thread'],
options['--remove-rRNA'], options['--max-bundle-frags'])
# convert assembly results
convert_assembly_gtf(out_dir, cufflinks_dir, options['--ref'],
options['--bb'], options['--chrom-size'])
def denovo(options):
# check output directory
out_dir = check_dir(options['<circ_dir>'])
# check tophat results
if options['--tophat-dir']:
tophat_dir = check_dir(options['--tophat-dir'])
else:
tophat_dir = check_dir(out_dir + '/tophat')
# prepare denovo directory
denovo_dir = '%s/denovo' % out_dir
create_dir(denovo_dir)
# combine ref files
cufflinks_ref_path = '%s/cufflinks/transcripts_ref.txt' % out_dir
if os.path.isfile(cufflinks_ref_path):
print('Combine %s with %s to create a new ref file!' %
(options['--ref'], cufflinks_ref_path))
ref_path = '%s/combined_ref.txt' % denovo_dir
new_ref_f = open(ref_path, 'w')
with open(cufflinks_ref_path, 'r') as cuff_ref:
for line in cuff_ref:
if line.startswith('CUFF'): # only import novel isoforms
new_ref_f.write(line)
new_ref_f.write(open(options['--ref'], 'r').read())
new_ref_f.close()
else:
print('Warning: no cufflinks directory under %s!' % out_dir)
print('Please run CIRCexplorer2 assembly before this step!')
ref_path = options['--ref']
# annotate fusion junctions
annotate_fusion(ref_path, denovo_dir, denovo_flag=1)
# fix fusion juncrions
fix_fusion(ref_path,
options['--genome'],
denovo_dir,
options['--no-fix'],
denovo_flag=1)
# extract novel circRNAs
extract_novel_circ(denovo_dir, options['--ref'])
if options['--as']:
if options['--pAplus'] and os.path.isdir(options['--pAplus']):
pAplus_dir = os.path.abspath(options['--pAplus'])
else:
sys.exit('You should offer --pAplus option in --as mode!')
if not options['--as-type'] or options['--as-type'] == 'CE':
# extract cassette exons
extract_cassette_exon(denovo_dir, tophat_dir, pAplus_dir,
options['--rpkm'])
if not options['--as-type'] or options['--as-type'] == 'RI':
# extract retained introns
extract_retained_intron(denovo_dir, tophat_dir, pAplus_dir)
if not options['--as-type'] or options['--as-type'] == 'ASS':
# characterize A5SS and A3SS
parse_splice_site(denovo_dir, tophat_dir, pAplus_dir)
def denovo(options):
# check output directory
out_dir = check_dir(options['<circ_dir>'])
# check tophat results
if options['--tophat-dir']:
tophat_dir = check_dir(options['--tophat-dir'])
else:
tophat_dir = check_dir(out_dir + '/tophat')
# prepare denovo directory
denovo_dir = '%s/denovo' % out_dir
create_dir(denovo_dir)
# combine ref files
cufflinks_ref_path = '%s/cufflinks/transcripts_ref.txt' % out_dir
if os.path.isfile(cufflinks_ref_path):
print('Combine %s with %s to create a new ref file!' %
(options['--ref'], cufflinks_ref_path))
ref_path = '%s/combined_ref.txt' % denovo_dir
new_ref_f = open(ref_path, 'w')
with open(cufflinks_ref_path, 'r') as cuff_ref:
for line in cuff_ref:
if line.startswith('CUFF'): # only import novel isoforms
new_ref_f.write(line)
new_ref_f.write(open(options['--ref'], 'r').read())
new_ref_f.close()
else:
print('Warning: no cufflinks directory under %s!' % out_dir)
print('Please run CIRCexplorer2 assembly before this step!')
ref_path = options['--ref']
# annotate fusion junctions
annotate_fusion(ref_path, denovo_dir, 1)
# fix fusion juncrions
fix_fusion(ref_path, options['--genome'], denovo_dir,
options['--no-fix'], 1)
# extract novel circRNAs
extract_novel_circ(denovo_dir, options['--ref'])
if options['--as']:
if options['--pAplus'] and os.path.isdir(options['--pAplus']):
pAplus_dir = os.path.abspath(options['--pAplus'])
else:
sys.exit('You should offer --pAplus option in --as mode!')
if not options['--as-type'] or options['--as-type'] == 'CE':
# extract cassette exons
extract_cassette_exon(denovo_dir, tophat_dir, pAplus_dir,
options['--rpkm'])
if not options['--as-type'] or options['--as-type'] == 'RI':
# extract retained introns
extract_retained_intron(denovo_dir, tophat_dir, pAplus_dir)
if not options['--as-type'] or options['--as-type'] == 'ASS':
# characterize A5SS and A3SS
parse_splice_site(denovo_dir, tophat_dir, pAplus_dir)
def annotate(options):
# check output directory
out_dir = check_dir(options['<circ_dir>'])
# prepare annotate directory
annotate_dir = '%s/annotate' % out_dir
create_dir(annotate_dir)
# annotate fusion junctions
annotate_fusion(options['--ref'], annotate_dir)
# fix fusion juncrions
fix_fusion(options['--ref'], options['--genome'], annotate_dir,
options['--no-fix'])
def annotate(options):
# check output directory
out_dir = check_dir(options['<circ_dir>'])
# prepare annotate directory
annotate_dir = '%s/annotate' % out_dir
create_dir(annotate_dir)
# annotate fusion junctions
annotate_fusion(options['--ref'], annotate_dir)
# fix fusion juncrions
fix_fusion(options['--ref'], options['--genome'], annotate_dir,
options['--no-fix'])