fileflag = 0;
indir, script_outdir = gwctcore.getOutdir(ins, "run_codeml", starttime, outdir_suffix);
outdir = os.path.join(script_outdir, "codeml_out");
filelist = os.listdir(indir);
if aopt == 1:
ancdir = os.path.join(script_outdir, "anc_seqs_fa");
print gwctcore.getTime() + " | Creating main output directory:\t" + script_outdir;
os.system("mkdir " + script_outdir);
logfilename = os.path.join(script_outdir, "run_codeml.log");
logfile = open(logfilename, "w");
logfile.write("");
logfile.close();
gwctcore.logCheck(l, logfilename, "=======================================================================");
gwctcore.logCheck(l, logfilename, "\t\t\tRunning codeml");
gwctcore.logCheck(l, logfilename, "\t\t\t" + gwctcore.getDateTime());
if fileflag == 1:
gwctcore.logCheck(l, logfilename, "INPUT | Making tree from file:\t\t" + ins);
else:
gwctcore.logCheck(l, logfilename, "INPUT | Running codeml on all files in:\t" + indir);
gwctcore.logCheck(l, logfilename, "INFO | PAML path set to:\t\t\t" + ppath);
if treefile != "":
gwctcore.logCheck(l, logfilename, "INFO | Using tree from file:\t\t" + treefile);
else:
gwctcore.logCheck(l, logfilename, "INFO | No tree file specified. codeml will infer a tree for each gene.");
if prune == 1:
gwctcore.logCheck(l, logfilename, "INFO | Pruning the tree for each gene.");
if bsopt == 0:
gwctcore.logCheck(l, logfilename, "INFO | Not doing branch-site test.");
targets = targets.split(",");
indir, script_outdir = gwctcore.getOutdir(indir, "gwct", starttime, suffix);
print "# " + gwctcore.getTime() + " | Creating main output directory:\t" + script_outdir;
os.system("mkdir " + script_outdir);
logfilename = os.path.join(script_outdir, "gwct.log");
logfile = open(logfilename, "w");
logfile.write("");
logfile.close();
uniq_header = "#\tChromosome\tGeneID\tAlignLen\tPosition\tBackgroundAlleles\tTargetAlleles\n";
l = 1;
sp = 65;
gwctcore.logCheck(l, logfilename, "# =======================================================================");
gwctcore.logCheck(l, logfilename, "#\tGWCT: Genome-Wide Convergence Tester -- Unique substitutions");
gwctcore.logCheck(l, logfilename, "#\t\t\t" + gwctcore.getDateTime());
gwctcore.logCheck(l, logfilename, gwctcore.spacedOut("# INPUT | Using sequences in:", sp) + indir);
gwctcore.logCheck(l, logfilename, gwctcore.spacedOut("# INPUT | Target species:", sp) + ",".join(targets));
gwctcore.logCheck(l, logfilename, gwctcore.spacedOut("# OUTPUT | Output directory created within input directory:", sp) + script_outdir);
gwctcore.logCheck(l, logfilename, "# ---------------------------------------------");
gwctcore.logCheck(2, logfilename, uniq_header);
################################
print "# " + gwctcore.getTime() + " | Counting unique substitutions....";
filelist = os.listdir(indir);
numfiles = len(filelist);
numbars = 0;
donepercent = [];
i = 0;
indir, script_outdir = gwctcore.getOutdir(indir, "gwct", starttime, suffix)
print "# " + gwctcore.getTime(
) + " | Creating main output directory:\t" + script_outdir
os.system("mkdir " + script_outdir)
logfilename = os.path.join(script_outdir, "gwct.log")
logfile = open(logfilename, "w")
logfile.write("")
logfile.close()
uniq_header = "#\tChromosome\tGeneID\tAlignLen\tPosition\tBackgroundAlleles\tTargetAlleles\n"
l = 1
sp = 65
gwctcore.logCheck(
l, logfilename,
"# ======================================================================="
)
gwctcore.logCheck(
l, logfilename,
"#\tGWCT: Genome-Wide Convergence Tester -- Unique substitutions")
gwctcore.logCheck(l, logfilename, "#\t\t\t" + gwctcore.getDateTime())
gwctcore.logCheck(
l, logfilename,
gwctcore.spacedOut("# INPUT | Using sequences in:", sp) + indir)
gwctcore.logCheck(
l, logfilename,
gwctcore.spacedOut("# INPUT | Target species:", sp) + ",".join(targets))
gwctcore.logCheck(
l, logfilename,
gwctcore.spacedOut(
"# OUTPUT | Output directory created within input directory:", sp) +
filelist = os.listdir(indir);
num_files = len(filelist);
print gwctcore.getTime() + " | Creating main output directory:\t" + script_outdir;
os.system("mkdir " + script_outdir);
logfilename = os.path.join(script_outdir, "gwct.log");
logfile = open(logfilename, "w");
logfile.write("");
logfile.close();
main_header = "# Chromosome\tGeneID\tAlignLen\tPosition\tAncAlleles\tTargetAlleles\n";
uniq_header = "# Chromosome\tGeneID\tAlignLen\tPosition\tBackgroundAlleles\tTargetAlleles\n";
l = 1;
sp = 65;
gwctcore.logCheck(l, logfilename, "# =======================================================================");
gwctcore.logCheck(l, logfilename, "#\t\tGWCT: Genome-Wide Convergence Tester");
gwctcore.logCheck(l, logfilename, "#\t\t\t" + gwctcore.getDateTime());
gwctcore.logCheck(l, logfilename, gwctcore.spacedOut("# INPUT | Using ancestral reconstructions in:", sp) + indir);
if orig_targets != "":
orig_targets = orig_targets.split(" ");
for t in xrange(len(orig_targets)):
if orig_targets[t].find(",") != -1:
orig_targets[t] = orig_targets[t].split(",");
if prob_thresh == 0:
convfilename = os.path.join(script_outdir, "conv_sites.txt");
divfilename = os.path.join(script_outdir, "div_sites.txt");
elif prob_thresh != 0:
convfilename = os.path.join(script_outdir, "conv_sites_" + str(prob_thresh) + ".txt");
divfilename = os.path.join(script_outdir, "div_sites_" + str(prob_thresh) + ".txt");
gwctcore.filePrep(convfilename, main_header);
print gwctcore.getTime(
) + " | Creating main output directory:\t" + script_outdir
os.system("mkdir " + script_outdir)
logfilename = os.path.join(script_outdir, "gwct.log")
logfile = open(logfilename, "w")
logfile.write("")
logfile.close()
main_header = "# Chromosome\tGeneID\tAlignLen\tPosition\tAncAlleles\tTargetAlleles\n"
uniq_header = "# Chromosome\tGeneID\tAlignLen\tPosition\tBackgroundAlleles\tTargetAlleles\n"
l = 1
sp = 65
gwctcore.logCheck(
l, logfilename,
"# ======================================================================="
)
gwctcore.logCheck(l, logfilename, "#\t\tGWCT: Genome-Wide Convergence Tester")
gwctcore.logCheck(l, logfilename, "#\t\t\t" + gwctcore.getDateTime())
gwctcore.logCheck(
l, logfilename,
gwctcore.spacedOut("# INPUT | Using ancestral reconstructions in:", sp) +
indir)
if orig_targets != "":
orig_targets = orig_targets.split(" ")
for t in xrange(len(orig_targets)):
if orig_targets[t].find(",") != -1:
orig_targets[t] = orig_targets[t].split(",")
if prob_thresh == 0:
convfilename = os.path.join(script_outdir, "conv_sites.txt")
divfilename = os.path.join(script_outdir, "div_sites.txt")
