def __main__():
print "Restricted to species:",sys.argv[3]
input_filename = sys.argv[1]
output_filename = sys.argv[2]
species = sys.argv[3].split(',')
partial = sys.argv[4]
num_species = len(species)
file_in = open(input_filename, 'r')
maf_reader = maf.Reader( file_in )
file_out = open(output_filename, 'w')
block_num=-1
for i, m in enumerate( maf_reader ):
block_num += 1
if "None" not in species:
m = m.limit_to_species( species )
l = m.components
if len(l) < num_species and partial == "partial_disallowed": continue
for c in l:
spec,chrom = maf.src_split( c.src )
file_out.write(">"+c.src+"("+c.strand+"):"+str(c.start)+"-"+str(c.end)+"|"+spec+"_"+str(block_num)+"\n")
file_out.write(c.text+"\n")
file_out.write("\n")
file_in.close()
file_out.close()
def __main__():
print "Restricted to species:", sys.argv[3]
input_filename = sys.argv[1]
output_filename = sys.argv[2]
species = sys.argv[3].split(",")
partial = sys.argv[4]
num_species = len(species)
file_in = open(input_filename, "r")
try:
maf_reader = maf.Reader(file_in)
file_out = open(output_filename, "w")
for block_num, block in enumerate(maf_reader):
if "None" not in species:
block = block.limit_to_species(species)
if len(block.components) < num_species and partial == "partial_disallowed":
continue
for component in block.components:
spec, chrom = maf.src_split(component.src)
if not spec or not chrom:
spec = chrom = component.src
file_out.write("%s\n" % maf_utilities.get_fasta_header(component, suffix="%s_%i" % (spec, block_num)))
file_out.write("%s\n" % component.text)
file_out.write("\n")
file_in.close()
except Exception, e:
print >> sys.stderr, "Your MAF file appears to be malformed:", e
sys.exit()
def __main__():
print "Restricted to species:", sys.argv[3]
input_filename = sys.argv[1]
output_filename = sys.argv[2]
species = sys.argv[3].split(',')
partial = sys.argv[4]
num_species = len(species)
file_in = open(input_filename, 'r')
maf_reader = maf.Reader(file_in)
file_out = open(output_filename, 'w')
block_num = -1
for i, m in enumerate(maf_reader):
block_num += 1
if "None" not in species:
m = m.limit_to_species(species)
l = m.components
if len(l) < num_species and partial == "partial_disallowed": continue
for c in l:
spec, chrom = maf.src_split(c.src)
file_out.write(">" + c.src + "(" + c.strand + "):" + str(c.start) +
"-" + str(c.end) + "|" + spec + "_" +
str(block_num) + "\n")
file_out.write(c.text + "\n")
file_out.write("\n")
file_in.close()
file_out.close()
def get_available_species(input_filename):
try:
rval = []
species = {}
file_in = open(input_filename, 'r')
maf_reader = maf.Reader(file_in)
for i, m in enumerate(maf_reader):
l = m.components
for c in l:
spec, chrom = maf.src_split(c.src)
species[spec] = spec
file_in.close()
species = species.keys()
species.sort()
file_in = open(input_filename, 'r')
maf_reader = maf.Reader(file_in)
species_sequence = {}
for s in species:
species_sequence[s] = []
for m in maf_reader:
for s in species:
c = m.get_component_by_src_start(s)
if c: species_sequence[s].append(c.text)
else: species_sequence[s].append("-" * m.text_size)
file_in.close()
rval.append(("Include all species.", 'None', False))
for spec in species:
species_sequence[spec] = "".join(species_sequence[spec])
rval.append((spec + ": " + str(
len(species_sequence[spec]) -
species_sequence[spec].count("-")) + " nongap, " +
str(len(species_sequence[spec])) + " total bases",
spec, False))
return rval
except:
return [(
"Include all species. <i>If you want to limit your conversion to a set of species, you must wait for the MAF file to be created.</i>",
'None', True)]
def get_species_names(input_filename):
species = {}
file_in = open(input_filename, 'r')
maf_reader = maf.Reader(file_in)
for i, m in enumerate(maf_reader):
l = m.components
for c in l:
spec, chrom = maf.src_split(c.src)
try:
if chrom not in species[spec]:
species[spec].append(chrom)
except:
species[spec] = [chrom]
file_in.close()
return species
def get_species_names( input_filename ):
species={}
file_in = open(input_filename, 'r')
maf_reader = maf.Reader( file_in )
for i, m in enumerate( maf_reader ):
l = m.components
for c in l:
spec, chrom = maf.src_split( c.src )
try:
if chrom not in species[spec]:
species[spec].append(chrom)
except:
species[spec] = [chrom]
file_in.close()
return species
def get_species(maf_filename):
try:
species={}
file_in = open(maf_filename, 'r')
maf_reader = maf.Reader( file_in )
for i, m in enumerate( maf_reader ):
l = m.components
for c in l:
spec,chrom = maf.src_split( c.src )
species[spec]=spec
file_in.close()
species = species.keys()
species.sort()
return species
except:
return []
def get_species(maf_filename):
try:
species = {}
file_in = open(maf_filename, 'r')
maf_reader = maf.Reader(file_in)
for i, m in enumerate(maf_reader):
l = m.components
for c in l:
spec, chrom = maf.src_split(c.src)
species[spec] = spec
file_in.close()
species = species.keys()
species.sort()
return species
except:
return []
def get_available_species( input_filename ):
try:
rval = []
species={}
file_in = open(input_filename, 'r')
maf_reader = maf.Reader( file_in )
for i, m in enumerate( maf_reader ):
l = m.components
for c in l:
spec,chrom = maf.src_split( c.src )
species[spec]=spec
file_in.close()
species = species.keys()
species.sort()
file_in = open(input_filename, 'r')
maf_reader = maf.Reader( file_in )
species_sequence={}
for s in species: species_sequence[s] = []
for m in maf_reader:
for s in species:
c = m.get_component_by_src_start( s )
if c: species_sequence[s].append( c.text )
else: species_sequence[s].append( "-" * m.text_size )
file_in.close()
rval.append( ("Include all species.",'None',False) )
for spec in species:
species_sequence[spec] = "".join(species_sequence[spec])
rval.append( (spec + ": "+str(len(species_sequence[spec]) - species_sequence[spec].count("-"))+" nongap, "+ str(len(species_sequence[spec])) + " total bases",spec,False) )
return rval
except:
return [("Include all species. <i>If you want to limit your conversion to a set of species, you must wait for the MAF file to be created.</i>",'None',True)]
def __main__():
input_filename = sys.argv[1]
output_filename = sys.argv[2]
# where to store files that become additional output
database_tmp_dir = sys.argv[5]
species = sys.argv[3].split(',')
partial = sys.argv[4]
output_id = sys.argv[6]
out_files = {}
primary_spec = None
if "None" in species:
species = set()
try:
for i, m in enumerate(maf.Reader(open(input_filename, 'r'))):
for c in m.components:
spec, chrom = maf.src_split(c.src)
if not spec or not chrom:
spec = chrom = c.src
species.add(spec)
except:
print("Invalid MAF file specified", file=sys.stderr)
return
if "?" in species:
print("Invalid dbkey specified", file=sys.stderr)
return
for i, spec in enumerate(species):
if i == 0:
out_files[spec] = open(output_filename, 'w')
primary_spec = spec
else:
out_files[spec] = open(
os.path.join(
database_tmp_dir,
'primary_%s_%s_visible_bed_%s' % (output_id, spec, spec)),
'wb+')
num_species = len(species)
print("Restricted to species:", ",".join(species))
file_in = open(input_filename, 'r')
maf_reader = maf.Reader(file_in)
block_num = -1
for i, m in enumerate(maf_reader):
block_num += 1
if "None" not in species:
m = m.limit_to_species(species)
l = m.components
if len(l) < num_species and partial == "partial_disallowed":
continue
for c in l:
spec, chrom = maf.src_split(c.src)
if not spec or not chrom:
spec = chrom = c.src
if spec not in out_files.keys():
out_files[spec] = open(
os.path.join(
database_tmp_dir, 'primary_%s_%s_visible_bed_%s' %
(output_id, spec, spec)), 'wb+')
if c.strand == "-":
out_files[spec].write(chrom + "\t" + str(c.src_size - c.end) +
"\t" + str(c.src_size - c.start) + "\t" +
spec + "_" + str(block_num) + "\t" +
"0\t" + c.strand + "\n")
else:
out_files[spec].write(chrom + "\t" + str(c.start) + "\t" +
str(c.end) + "\t" + spec + "_" +
str(block_num) + "\t" + "0\t" +
c.strand + "\n")
file_in.close()
for file_out in out_files.keys():
out_files[file_out].close()
print("#FILE1_DBKEY\t%s" % (primary_spec))
def __main__():
input_filename = sys.argv[1]
output_filename = sys.argv[2]
# where to store files that become additional output
database_tmp_dir = sys.argv[5]
species = sys.argv[3].split(',')
partial = sys.argv[4]
output_id = sys.argv[6]
out_files = {}
primary_spec = None
if "None" in species:
species = OrderedDict()
try:
for i, m in enumerate(maf.Reader(open(input_filename, 'r'))):
for c in m.components:
spec, chrom = maf.src_split(c.src)
if not spec or not chrom:
spec = c.src
species[spec] = None
species = species.keys()
except Exception:
print("Invalid MAF file specified", file=sys.stderr)
return
if "?" in species:
print("Invalid dbkey specified", file=sys.stderr)
return
for i, spec in enumerate(species):
if i == 0:
out_files[spec] = open(output_filename, 'w')
primary_spec = spec
else:
out_files[spec] = open(os.path.join(database_tmp_dir, 'primary_%s_%s_visible_bed_%s' % (output_id, spec, spec)), 'w+')
num_species = len(species)
print("Restricted to species:", ",".join(species))
with open(input_filename, 'r') as file_in:
maf_reader = maf.Reader(file_in)
block_num = -1
for i, m in enumerate(maf_reader):
block_num += 1
if "None" not in species:
m = m.limit_to_species(species)
l = m.components
if len(l) < num_species and partial == "partial_disallowed":
continue
for c in l:
spec, chrom = maf.src_split(c.src)
if not spec or not chrom:
spec = chrom = c.src
if spec not in out_files.keys():
out_files[spec] = open(os.path.join(database_tmp_dir, 'primary_%s_%s_visible_bed_%s' % (output_id, spec, spec)), 'wb+')
if c.strand == "-":
out_files[spec].write(chrom + "\t" + str(c.src_size - c.end) + "\t" + str(c.src_size - c.start) + "\t" + spec + "_" + str(block_num) + "\t" + "0\t" + c.strand + "\n")
else:
out_files[spec].write(chrom + "\t" + str(c.start) + "\t" + str(c.end) + "\t" + spec + "_" + str(block_num) + "\t" + "0\t" + c.strand + "\n")
for file_out in out_files.keys():
out_files[file_out].close()
print("#FILE1_DBKEY\t%s" % (primary_spec))
def __main__():
input_filename = sys.argv[1]
output_filename = sys.argv[2]
#where to store files that become additional output
database_tmp_dir = sys.argv[5]
species = sys.argv[3].split(',')
partial = sys.argv[4]
out_files = {}
primary_spec = None
if "None" in species:
species = {}
try:
for i, m in enumerate( maf.Reader( open( input_filename, 'r' ) ) ):
for c in m.components:
spec,chrom = maf.src_split( c.src )
if not spec or not chrom:
spec = chrom = c.src
species[spec] = ""
species = species.keys()
except:
print >>sys.stderr, "Invalid MAF file specified"
return
if "?" in species:
print >>sys.stderr, "Invalid dbkey specified"
return
for i in range( 0, len( species ) ):
spec = species[i]
if i == 0:
out_files[spec] = open( output_filename, 'w' )
primary_spec = spec
else:
out_files[spec] = tempfile.NamedTemporaryFile( mode = 'w', dir = database_tmp_dir, suffix = '.maf_to_bed' )
filename = out_files[spec].name
out_files[spec].close()
out_files[spec] = open( filename, 'w' )
num_species = len( species )
print "Restricted to species:", ",".join( species )
file_in = open( input_filename, 'r' )
maf_reader = maf.Reader( file_in )
block_num = -1
for i, m in enumerate( maf_reader ):
block_num += 1
if "None" not in species:
m = m.limit_to_species( species )
l = m.components
if len(l) < num_species and partial == "partial_disallowed": continue
for c in l:
spec,chrom = maf.src_split( c.src )
if not spec or not chrom:
spec = chrom = c.src
if spec not in out_files.keys():
out_files[spec] = tempfile.NamedTemporaryFile( mode='w', dir = database_tmp_dir, suffix = '.maf_to_bed' )
filename = out_files[spec].name
out_files[spec].close()
out_files[spec] = open( filename, 'w' )
if c.strand == "-":
out_files[spec].write( chrom + "\t" + str( c.src_size - c.end ) + "\t" + str( c.src_size - c.start ) + "\t" + spec + "_" + str( block_num ) + "\t" + "0\t" + c.strand + "\n" )
else:
out_files[spec].write( chrom + "\t" + str( c.start ) + "\t" + str( c.end ) + "\t" + spec + "_" + str( block_num ) + "\t" + "0\t" + c.strand + "\n" )
file_in.close()
for file_out in out_files.keys():
out_files[file_out].close()
for spec in out_files.keys():
if spec != primary_spec:
print "#FILE\t" + spec + "\t" + os.path.join( database_tmp_dir, os.path.split( out_files[spec].name )[1] )
else:
print "#FILE1\t" + spec + "\t" + out_files[spec].name
def __main__():
input_filename = sys.argv[1]
output_filename = sys.argv[2]
#where to store files that become additional output
database_tmp_dir = sys.argv[5]
species = sys.argv[3].split(',')
partial = sys.argv[4]
out_files = {}
primary_spec = None
if "None" in species:
species = {}
try:
for i, m in enumerate(maf.Reader(open(input_filename, 'r'))):
for c in m.components:
spec, chrom = maf.src_split(c.src)
if not spec or not chrom:
spec = chrom = c.src
species[spec] = ""
species = species.keys()
except:
print >> sys.stderr, "Invalid MAF file specified"
return
if "?" in species:
print >> sys.stderr, "Invalid dbkey specified"
return
for i in range(0, len(species)):
spec = species[i]
if i == 0:
out_files[spec] = open(output_filename, 'w')
primary_spec = spec
else:
out_files[spec] = tempfile.NamedTemporaryFile(mode='w',
dir=database_tmp_dir,
suffix='.maf_to_bed')
filename = out_files[spec].name
out_files[spec].close()
out_files[spec] = open(filename, 'w')
num_species = len(species)
print "Restricted to species:", ",".join(species)
file_in = open(input_filename, 'r')
maf_reader = maf.Reader(file_in)
block_num = -1
for i, m in enumerate(maf_reader):
block_num += 1
if "None" not in species:
m = m.limit_to_species(species)
l = m.components
if len(l) < num_species and partial == "partial_disallowed": continue
for c in l:
spec, chrom = maf.src_split(c.src)
if not spec or not chrom:
spec = chrom = c.src
if spec not in out_files.keys():
out_files[spec] = tempfile.NamedTemporaryFile(
mode='w', dir=database_tmp_dir, suffix='.maf_to_bed')
filename = out_files[spec].name
out_files[spec].close()
out_files[spec] = open(filename, 'w')
if c.strand == "-":
out_files[spec].write(chrom + "\t" + str(c.src_size - c.end) +
"\t" + str(c.src_size - c.start) + "\t" +
spec + "_" + str(block_num) + "\t" +
"0\t" + c.strand + "\n")
else:
out_files[spec].write(chrom + "\t" + str(c.start) + "\t" +
str(c.end) + "\t" + spec + "_" +
str(block_num) + "\t" + "0\t" +
c.strand + "\n")
file_in.close()
for file_out in out_files.keys():
out_files[file_out].close()
for spec in out_files.keys():
if spec != primary_spec:
print "#FILE\t" + spec + "\t" + os.path.join(
database_tmp_dir,
os.path.split(out_files[spec].name)[1])
else:
print "#FILE1\t" + spec + "\t" + out_files[spec].name