else:
if not ins.endswith("/"):
ins = ins + "/"
filelist = os.listdir(ins)
tot_pos = 0
numlines = len(filelist)
numbars = 0
donepercent = []
i = 0
for each in filelist:
if disp_file == 0:
numbars, donepercent = core.loadingBar(i, numlines, donepercent,
numbars)
elif disp_file == 1:
print "----------"
print each
print "Sequence\tLength"
i = i + 1
if each.find(".fa") == -1:
continue
specpos = 0
infilename = ins + each
inseqs = core.fastaGetDict(infilename)
if a.find(".fa") != -1:
num_aligns = num_aligns + 1;
if v == 0:
gb_logfile = outdir + "gblocks.log";
core.logCheck(l, logfilename, core.getTime() + " | Runnning GBlocks on " + str(num_aligns) + " alignments...");
acc_mask = 0;
i = 0;
numbars = 0;
donepercent = [];
for each in filelist:
if v == 0 and fileflag == 0:
numbars, donepercent = core.loadingBar(i, len(filelist), donepercent, numbars);
i = i + 1;
if each.find(".fa") == -1:
continue;
if fileflag == 1:
infilename = each;
if each.find("/") != -1:
gb_outfile = each[each.rfind("/")+1:each.index(".fa")] + "-gb.fa";
else:
gb_outfile = each[:each.index(".fa")] + "-gb.fa";
else:
infilename = indir + each;
gb_outfile = each[:each.index(".fa")] + "-gb.fa";
if v == 0:
gb_logfile = outdir + "gblocks.log"
core.logCheck(
l, logfilename,
core.getTime() + " | Runnning GBlocks on " + str(num_aligns) +
" alignments...")
acc_mask = 0
i = 0
numbars = 0
donepercent = []
for each in filelist:
if v == 0 and fileflag == 0:
numbars, donepercent = core.loadingBar(i, len(filelist), donepercent,
numbars)
i = i + 1
if each.find(".fa") == -1:
continue
if fileflag == 1:
infilename = each
if each.find("/") != -1:
gb_outfile = each[each.rfind("/") + 1:each.index(".fa")] + "-gb.fa"
else:
gb_outfile = each[:each.index(".fa")] + "-gb.fa"
else:
infilename = indir + each
gb_outfile = each[:each.index(".fa")] + "-gb.fa"
else:
if not ins.endswith("/"):
ins = ins + "/";
filelist = os.listdir(ins);
tot_pos = 0;
numlines = len(filelist);
numbars = 0;
donepercent = [];
i = 0;
for each in filelist:
if disp_file == 0:
numbars, donepercent = core.loadingBar(i, numlines, donepercent, numbars);
elif disp_file == 1:
print "----------";
print each;
print "Sequence\tLength";
i = i + 1;
if each.find(".fa") == -1:
continue;
specpos = 0;
infilename = ins + each;
inseqs = core.fastaGetDict(infilename);
print "# " + core.getTime() + " Retrieving key IDs in both files..."
comb_ids = list(set.intersection(*map(set, key_ids.values())))
num_ids = len(comb_ids)
print "# -------------------------------------"
print "# " + core.getTime(
) + " Retrieving lines with shared key IDs (all combinations)..."
i = 0
numbars = 0
donepercent = []
outfile = open(outfilename, "w")
for cid in comb_ids:
numbars, donepercent = core.loadingBar(i, num_ids, donepercent, numbars)
i += 1
cur_id = {}
for each in infiles:
cur_id[each] = []
for line in file_lines[each]:
if line[0] == cid:
cur_id[each].append(line)
id_vals = []
for each in cur_id:
id_vals.append(cur_id[each])
for b in itertools.product(*id_vals):
outline = b[0] + b[1][1:]
outline = "\t".join(outline) + "\n"
main_seq_dict[spec][new_title] = tmp_seq_dict[spec][title]
del tmp_seq_dict
print "# -------------------------------------"
count = core.getFileLen(infilename)
print "# " + core.getTime() + " Combining", count, "orthologs..."
i = 0
numbars = 0
donepercent = []
nonorth = 0
for line in open(infilename):
numbars, donepercent = core.loadingBar(i, count, donepercent, numbars)
tmpline = line.replace("\n", "").split("\t")
if i == 0:
numspec = len(tmpline)
i = i + 1
if len(tmpline) != numspec:
print line
nonorth = nonroth + 1
continue
sn = 0
finalseqs = {}
for gid in tmpline:
if sn == 0:
fcritcount = 0;
ocritcount = 0;
zcount = 0;
noln_genes = [];
i = 0;
numfiles = len(filelist);
numbars = 0;
donepercent = [];
print "Performing LRT on PAML output to test for positive selection...";
for each in filelist:
if each.find(".fa") == -1:
continue;
numbars, donepercent = core.loadingBar(i, numfiles, donepercent, numbars);
i = i + 1;
# gid = each[:each.index(".fa")];
altfilename = os.path.join(altdir + "codeml_out", each, each + ".out");
nullfilename = os.path.join(nulldir + "codeml_out", each, each + ".out");
#Reading the alt file...
altfile = open(altfilename, "r");
altlines = altfile.readlines();
altfile.close();
altlnflag = 0;
for alt in altlines:
if alt[:3] == "lnL":
if alt.find("nan") != -1:
fcritcount = 0
ocritcount = 0
zcount = 0
noln_genes = []
i = 0
numfiles = len(filelist)
numbars = 0
donepercent = []
print "Performing LRT on PAML output to test for positive selection..."
for each in filelist:
if each.find(".fa") == -1:
continue
numbars, donepercent = core.loadingBar(i, numfiles, donepercent, numbars)
i = i + 1
gid = each[: each.index(".fa")]
altfilename = os.path.join(altdir + "codeml_out", gid, gid + ".out")
nullfilename = os.path.join(nulldir + "codeml_out", gid, gid + ".out")
# Reading the alt file...
altfile = open(altfilename, "r")
altlines = altfile.readlines()
altfile.close()
altlnflag = 0
for alt in altlines:
if alt[:3] == "lnL":
if alt.find("nan") != -1:
