def run_sorting(blat_psl, results) :
print 'blat_psl is' ,blat_psl
utils.check_existence_or_raise(blat_psl)
sorted_psl = os.path.join(results,utils.get_name(blat_psl)+'.sorted.psl')
tmp = utils.atomic_tmp_file(sorted_psl)
params = ['sort', '-k 10,10', blat_psl, '>', tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp, sorted_psl)
sorted_psl_only_entries = os.path.join(results,utils.get_name(blat_psl)+'.sorted.only_entries.psl')
tmp = utils.atomic_tmp_file(sorted_psl_only_entries)
subprocess.call('tail -n +5 '+ sorted_psl + '| head -n -1 > ' + tmp, shell=True)
utils.atomic_install(tmp, sorted_psl_only_entries)
return sorted_psl_only_entries
def run_faToBit(fasta, path_to_results):
utils.check_existence_or_raise(fasta)
twobit = os.path.join(path_to_results, utils.get_name(fasta) + '.2bit')
tmp = utils.atomic_tmp_file(twobit)
params = ['faToTwoBit', '-noMask', fasta, tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp, twobit)
return twobit
def run_halLiftover(hal,specie_source,genome_bed_path,specie_target,output_prefix) :
bed = output_prefix+'.bed'
tmp=utils.atomic_tmp_file(bed)
params = ['halLiftover --inMemory',hal,specie_source,genome_bed_path,specie_target,tmp]
params=' '.join(params)
subprocess.call(params, shell=True)
utils.atomic_install(tmp,bed)
return bed
def run_blat(fasta, ncrna, results):
utils.check_existence_or_raise(fasta)
blat_psl = os.path.join(results, utils.get_name(fasta) + '.psl')
tmp = utils.atomic_tmp_file(blat_psl)
params = ['blat', "-q=rna", fasta, ncrna, tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp, blat_psl)
return blat_psl
def run_pslToBigBed(blat_psl, sizes, results):
utils.check_existence_or_raise(blat_psl)
blat_bed=os.path.join(results,utils.get_name(blat_psl)+'.bed')
tmp = utils.atomic_tmp_file(blat_bed)
params = ['pslToBed', blat_psl, tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp, blat_bed)
sorted_bed = os.path.join(results,utils.get_name(blat_psl)+'.sorted.bed')
tmp = utils.atomic_tmp_file(sorted_bed)
params = ['bedSort', blat_bed, tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp, sorted_bed)
bigBed = os.path.join(results,utils.get_name(blat_psl)+'.bigBed')
tmp = utils.atomic_tmp_file(bigBed)
params = ['bedToBigBed', sorted_bed, sizes, tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp, bigBed)
def run_faToBit(fasta, path_to_results):
utils.check_existence_or_raise(fasta)
twobit = os.path.join(path_to_results, utils.get_name(fasta)+'.2bit')
tmp = utils.atomic_tmp_file(twobit)
params = ['faToTwoBit', '-noMask', fasta, tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp,twobit)
return twobit
def run_wigToBigWig(fasta, wig, sizes, path_to_results):
utils.check_existence_or_raise(wig)
bigWig=os.path.join(path_to_results, utils.get_name(fasta)+'.gc5Base'+'.bw')
tmp = utils.atomic_tmp_file(bigWig)
params = ['wigToBigWig', wig, sizes, tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp,bigWig)
return bigWig
def run_gcPercent(fasta, twobit, path_to_results):
utils.check_existence_or_raise(twobit)
wig=os.path.join(path_to_results, 'gc5Base.'+utils.get_name(twobit)+'.wig')
tmp = utils.atomic_tmp_file(wig)
params = ['hgGcPercent', '-wigOut', '-doGaps', '-win=5', '-file='+tmp, '-verbose=0', utils.get_name(fasta), path_to_results]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp,wig)
return wig
def run_wigToBigWig(fasta, wig, sizes, path_to_results):
utils.check_existence_or_raise(wig)
bigWig = os.path.join(path_to_results,
utils.get_name(fasta) + '.gc5Base' + '.bw')
tmp = utils.atomic_tmp_file(bigWig)
params = ['wigToBigWig', wig, sizes, tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp, bigWig)
return bigWig
def run_genomecov(bed, size):
genomecov = bed.split('.')
genomecov.insert(1,'genomecov')
genomecov = '.'.join(genomecov)
tmp=utils.atomic_tmp_file(genomecov)
params = ['bedtools genomecov -i', bed, '-g', size, '>', tmp]
params=' '.join(params)
subprocess.call(params, shell=True)
utils.atomic_install(tmp,genomecov)
return genomecov
def run_sort(bed):
sorted = bed.split('.')
sorted.insert(1,'sorted')
sorted = '.'.join(sorted)
tmp=utils.atomic_tmp_file(sorted)
params = ['bedtools sort -i', bed, '>', tmp]
params=' '.join(params)
subprocess.call(params, shell=True)
utils.atomic_install(tmp,sorted)
return sorted
def run_gcPercent(fasta, twobit, path_to_results):
utils.check_existence_or_raise(twobit)
wig = os.path.join(path_to_results,
'gc5Base.' + utils.get_name(twobit) + '.wig')
tmp = utils.atomic_tmp_file(wig)
params = [
'hgGcPercent', '-wigOut', '-doGaps', '-win=5', '-file=' + tmp,
'-verbose=0',
utils.get_name(fasta), path_to_results
]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp, wig)
return wig
def run_filtering(sorted_psl, results) :
utils.check_existence_or_raise(sorted_psl)
stats = os.path.join(results,'filtering_stats')
utils.create_dir_if_not_exists(stats)
filtered_psl = os.path.join(results,utils.get_name(sorted_psl)+'.filtered.psl')
tmp = utils.atomic_tmp_file(filtered_psl)
params = ['pslCDnaFilter','-localNearBest=0.1 -ignoreNs -minCover=0.6 -repsAsMatch',
' -dropped='+os.path.join(stats,'dropped.out'), '-weirdOverlapped='+os.path.join(stats,'weird_overlapped.out'),
'-alignStats='+os.path.join(stats,'align_stats.out'), '-statsOut='+os.path.join(stats,'overall_stats.out'),
sorted_psl, tmp]
subprocess.call(" ".join(params), shell=True)
utils.atomic_install(tmp, filtered_psl)
return filtered_psl