def Align(self, s1, s2, result):
result.clear()
handle_tmpfile, filename_tmpfile = tempfile.mkstemp()
os.write(handle_tmpfile, ">s1\n%s\n" % (s1))
os.write(handle_tmpfile, ">s2\n%s\n" % (s2))
os.close(handle_tmpfile)
statement = string.join(("(", self.mEnvironment, self.mExecutable,
self.mOptions, filename_tmpfile, ")"), " ")
p = subprocess.Popen(statement,
shell=True,
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
close_fds=True)
(file_stdout, file_stdin, file_stderr) = (p.stdin, p.stdout, p.stderr)
file_stdin.close()
lines = file_stdout.readlines()
lines_stderr = file_stderr.readlines()
exit_code = file_stdout.close()
file_stderr.close()
if exit_code:
raise "Error while executing statement %s" % statement
r = None
for x in range(len(lines)):
if re.search("Alignment \(FASTA format\):", lines[x]):
r = Genomics.ParseFasta2Hash(lines[x + 2:])
break
if not r:
return None
a1 = r['s1']
a2 = r['s2']
x1 = 1
x2 = 1
for pos in range(len(a1)):
if a1[pos] in string.uppercase and a2[pos] in string.uppercase:
result.addPairExplicit(x1, x2, 0)
x1 += 1
x2 += 1
continue
if a1[pos] != "-":
x1 += 1
if a2[pos] != "-":
x2 += 1
os.remove(filename_tmpfile)
return result
def GetBlocks(self, s1, s2):
"""the strings have to be already aligned!!!"""
handle_tmpfile, filename_tmpfile = tempfile.mkstemp()
os.write(handle_tmpfile, ">s1\n%s\n" % (s1))
os.write(handle_tmpfile, ">s2\n%s\n" % (s2))
os.close(handle_tmpfile)
statement = " ".join("(", self.mEnvironment,
self.mExecutable % filename_tmpfile,
self.mOptions, ")")
p = subprocess.Popen(statement,
shell=True,
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
close_fds=True)
(file_stdout, file_stdin, file_stderr) = (p.stdin, p.stdout, p.stderr)
file_stdin.close()
lines = file_stdout.readlines()
lines_stderr = file_stderr.readlines()
exit_code = file_stdout.close()
file_stderr.close()
if exit_code:
raise ValueError("Error while executing statement %s" % statement)
if not os.path.exists(filename_tmpfile + "-gb"):
os.remove(filename_tmpfile)
return "", ""
lines = open(filename_tmpfile + "-gb").readlines()
r = Genomics.ParseFasta2Hash(lines)
if not r:
return "", ""
os.remove(filename_tmpfile)
os.remove(filename_tmpfile + "-gb")
os.remove(filename_tmpfile + "-gb.htm")
return r['s1'], r['s2']
