def align_fasta(infname, outfname, debug=False):
"""
Generate an alignment for the given fasta file.
Args:
infname (str): Path to fasta to be aligned.
outfname (str): Path to output fasta to be
"""
muscle_exec = {
"Windows": "niclassify/bin/muscle3.8.31_i86win32.exe",
"Linux": "niclassify/bin/muscle3.8.31_i86linux64",
"Darwin": "niclassify/bin/muscle3.8.31_i86darwin64"
}[PLATFORM]
alignment_call = MuscleCommandline(os.path.realpath(
os.path.join(MAIN_PATH, muscle_exec)),
input=os.path.realpath(infname),
out=os.path.realpath(outfname))
print(alignment_call.__str__())
if debug:
subprocess.run(alignment_call.__str__(),
creationflags=subprocess.CREATE_NEW_CONSOLE,
shell=True)
else:
subprocess.run(alignment_call.__str__(), shell=True)
r_script = os.path.realpath(
os.path.join(MAIN_PATH, "niclassify/core/scripts/trim_alignment.R"))
trim_call = [R_LOC, r_script, outfname, outfname]
if debug:
proc = subprocess.run(trim_call,
creationflags=subprocess.CREATE_NEW_CONSOLE,
env=os.environ.copy())
else:
proc = subprocess.run(trim_call, env=os.environ.copy())
if os.stat(outfname).st_size == 0:
raise ChildProcessError("Sequence Alignment Failed")
if proc.returncode != 0:
raise RScriptFailedError("R TrimAlignment failed")
def allign_fasta(filename = "filename",
extension_in = ".fasta",
extension_out = ".aln"):
"""
This function requires MUSCLE from http://www.drive5.com/muscle. The
main objective - read FASTA file with multiple records, find similar
sequences, save alingment of similar sequences to "filename.aln".
@param filename: FASTA file, which should be alligned.
@param extension_in: FASTA file type end, could be .fa or similar.
@param extension_out: Alignment file type: ".aln".
"""
from Bio.Align.Applications import MuscleCommandline
if filename == None:
return False;
if not os.path.exists(filename + extension_in):
return False;
cline = MuscleCommandline(input=filename + extension_in, out=filename + extension_out);
os.system(cline.__str__());
return True;
