def check_a_line(a_line):
if is_5col_bed(last_line):
return True
else:
if check_3col_bed:
if is_3col_bed(last_line):
warn("You input a 3 column bed file like this:\t\t%s" %
last_line[:50])
info("[3 Column to 5 Column] %s ==> %s " %
(suffix_5col(fname), fname))
threetofive(fname)
info(
"Use %s instead of %s as the input BED and run the executive Again"
% (suffix_5col(fname), fname))
else:
error(
"The input bed file %s has a wrong format!(3 column checking active)"
% fname)
print "Wrong Format:\t\t\t%s" % last_line[:50]
print "Right Format should look like:\t%s" % (
'chr1\t567577\t567578\tMACS_peak_1\t119.00')
print "Or the depreciate 3-column format like this:\t%s" % (
'chr1\t567577\t567578')
else:
error("The input bed file %s has a wrong format!" % fname)
print "Wrong Format:\t\t\t%s" % last_line[:50]
print "Right Format should look like:\t%s" % (
'chr1\t567577\t567578\tMACS_peak_1\t119.00')
return False
def check_fasta_dna(fname):
"""
Check if a file has the format of fasta
@type fname: str
@param fname: path of the file to be checked
@rtype: bool
@return: whether the file passed the fasta check
"""
if not check_common(fname, ".fa",
maxsize=10737418240): # 10G=10*1024^3=10737418240
return False
print fname
with open(fname) as fasta_f:
first_line = fasta_f.readline()
if not first_line[0] == ">":
error("The input fasta file %s has a wrong format!" % fname)
print "Wrong Format:\t\t\t%s" % first_line[:50]
print "Right Format should look like:\t%s" % (
'>chr1:1150372-1150572')
return False
second_line = fasta_f.readline()
fasta_pattern_scd = "[AGCTN]+"
if not re.search(fasta_pattern_scd, second_line):
error("The input fasta file %s has a wrong format!" % fname)
print "Wrong Format:\t\t\t%s" % second_line[:50]
print "Right Format should look like:\t%s" % ('NGGGCCATTCA')
return False
return True
def fetch_seq_record(fasta_file, alpha=_alphabet):
"""
Fetch the sequence's nucleotide order and position information from a fasta file
@type fasta_file: str
@param fasta_file: path of the XML file
@rtype: list
@return: sequence information
"""
if not check_fasta_dna(fasta_file):
error("fasta file validation failed")
sys.exit(1)
raw_seq_list = SeqIO.parse(fasta_file, "fasta", alpha)
return list(raw_seq_list)
def check_cmd(command_):
"""
Check whether a command can be run in shell
@type command_: str
@param command_: the command you want to check, for example, "awk"
@rtype: bool
@return: whether the command passed the check
"""
exit_code = subcall(command_, shell=True, stdout=-1, stderr=-1)
# Get the exit code without printing standard output
print("check command %s" % command_).center(30, "-")
if exit_code == 127: # when command not found, exit_code is 127
error("No such command as '%s'" % command_)
return False
else:
return True
def check_xml(fname):
"""
Check if a file has the format of xml
@type fname: str
@param fname: path of the file to be checked
@rtype: bool
@return: whether the file passed the xml check
"""
if not check_common(fname, ".xml",
maxsize=10485760): # 10M = 1024*1024*10 = 10485760
return False
xmltree = ElementTree()
try:
xmltree.parse(fname)
except:
error("Fail to parser the xml file.")
error(
"The input XML file %s has a wrong format, please check it again."
% fname)
return False
for pos in xmltree.findall("motif"):
#get key and set empty element
key = pos.get('id')
if not key:
error("No 'id' found for node, not a xml for motif information?")
return False
return True
def fetch_pssm_xml(xmlfile):
"""
Fetch the motif's pssm and other information from an XML file
@type xmlfile: str
@param xmlfile: path of the XML file
@rtype: dict
@return: motif information
"""
if not check_xml(xmlfile):
error("xml file validation failed")
sys.exit(1)
mp = MP.MotifParser()
mp.tag_list = ["dbd", "synonym", "description"]
mp.Parser(xmlfile)
for m_id in mp.motifs:
one_pssm = mp.motifs[m_id]['pssm']
for one_pos in one_pssm:
for i in range(0, 4):
if one_pos[i] == 0.0:
one_pos[i] += 0.00001
print "Found one"
return mp.motifs
def check_common(fname, suffix, maxsize=1073741824): # 1GB=1024^3=1073741824B
"""
Check if a file has the specified suffix and smaller than maxsize
@type fname: str
@param fname: path of the file to be checked
@type suffix: str
@param suffix: the suffix limit, if not matched, WARNING will appears, but WON'T fail in this check
@type maxsize: str
@param maxsize: the max limit of the file to be checked
@rtype: bool
@return: whether the file passed the check
"""
if not os.path.isfile(fname):
error("No such bed file: %s" % fname)
return False
if os.path.getsize(fname) > maxsize:
error("The input file %s is larger than maxsize:%d bytes!" %
(fname, maxsize))
return False
if not fname.endswith(suffix):
warn("Your input file %s doesn't have the suffix %s" % (fname, suffix))
return True