def finalize():
if not hasattr(GC,'final_sequences'): # GC.final_sequences[cn_node][t] = set of (label,seq) tuples
GC.final_sequences = {}
if GC.errorfree_sequence_file.lower().endswith('.gz'):
from gzip import open as gopen
lines = [l.decode().strip() for l in gopen(GC.errorfree_sequence_file)]
else:
lines = [l.strip() for l in open(GC.errorfree_sequence_file)]
lines = [l for l in lines if len(l) != 0]
if len(lines) == 0:
return
seqs = GC.parseFASTA(lines)
for ID,seq in seqs.items():
v,n,t = ID.split('|'); t = float(t)
if v == 'DUMMY':
continue
if n not in GC.final_sequences:
GC.final_sequences[n] = {}
if t not in GC.final_sequences[n]:
GC.final_sequences[n][t] = []
GC.final_sequences[n][t].append((v,seq))
def get_edge_list():
args = []
for arg in PANGEA_ARGS:
val = getattr(GC, arg)
if isinstance(val, str):
val = val.strip()
if len(val) == 0 and arg.split('_')[1].strip(
) == 'seed': # if no seed given, randomly generate
from random import randint
val = str(randint(0, 32767))
if len(val) != 0:
args.append("%s='%s'" % (arg.split('_')[1], val))
else:
args.append("%s=%s" % (arg.split('_')[1], val))
orig_dir = getcwd()
makedirs(PANGEA_path, exist_ok=True)
chdir(PANGEA_path)
f = open(PANGEA_command_script, 'w')
f.write("library(PANGEA.HIV.sim)\n")
f.write("outdir <- getwd()\n")
f.write("pipeline.args <- sim.regional.args(")
if GC.random_number_seed is not None:
f.write("seed=%d," % GC.random_number_seed)
GC.random_number_seed += 1
f.write(', '.join(args))
f.write(")\ncat(sim.regional(outdir, pipeline.args=pipeline.args))")
f.close()
try:
check_output([GC.Rscript_path, PANGEA_command_script],
stderr=open(devnull, 'w'))
except FileNotFoundError:
chdir(GC.START_DIR)
assert False, "Rscript executable was not found"
for script in glob('*.sh'):
break
script_str = open(script, 'r').read()
f = open(script, 'w')
f.write('#!/usr/bin/env bash\n%s' % script_str)
f.close()
check_output(['./%s' % script],
stderr=open('%s_output.log' % script, 'w'))
archive = None
for archive in glob('*_INTERNAL.zip'):
break
assert archive is not None, "PANGEA failed to run successfully"
z = ZipFile(archive, 'r')
internal = [
item for item in z.namelist()
if item.endswith('_SIMULATED_INTERNAL.R')
][0]
f = open(internal, 'wb')
f.write(z.read(internal))
f.close()
f = open(PANGEA_trans_net_script, 'w')
f.write("library(PANGEA.HIV.sim)\n")
f.write("load('%s')\n" % internal)
f.write("trans <- df.trms[,c('IDTR','IDREC','TIME_TR')]\n")
f.write(
"write.table(trans[order(trans$TIME_TR),], file='%s', append=FALSE, sep='\\t', row.names=FALSE, col.names=FALSE, quote=FALSE)"
% PANGEA_trans_file)
f.close()
check_output([GC.Rscript_path, PANGEA_trans_net_script],
stderr=open(devnull, 'w'))
GC.PANGEA_TRANSMISSION_NETWORK = [
i.strip().split() for i in open(PANGEA_trans_file)
if len(i.strip()) > 0
]
chdir(orig_dir)
for archive in glob('%s/*_SIMULATED_SEQ.zip' % PANGEA_path):
break
z = ZipFile(archive, 'r')
fasta_files = [item for item in z.namelist() if item.endswith('.fa')]
f = gopen("error_free_files/sequence_data.fasta.gz", 'wb', 9)
for fasta in fasta_files:
ending = '_%s.fasta' % fasta.split('_')[-1].split('.')[0]
seqs = GC.parseFASTA(z.read(fasta).decode('ascii').splitlines())
for seqID in seqs:
f.write(('>%s\n%s\n' % (seqID, seqs[seqID])).encode())
f.write(b'\n')
f.close()
archive = None
for archive in glob('%s/*_SIMULATED_TREE.zip' % PANGEA_path):
break
assert archive is not None, "PANGEA failed to run successfully"
z = ZipFile(archive, 'r')
trees = [item for item in z.namelist() if item.endswith('.newick')]
for tree in trees:
f = gopen("error_free_files/phylogenetic_trees/%s.gz" % tree, 'wb',
9)
to_write = z.read(tree)
if isinstance(to_write, bytes):
f.write(to_write)
else:
f.write(to_write.encode())
f.write(b'\n')
f.close()
return []