def compress_vcf(args, myvcf):
"""Compress a vcf file with bgzip and tabix"""
# make sure a zipped version of the file doesnt already exist
cmd = "rm -f " + myvcf + ".bgz"
whichcmd(cmd, args, 0)
cmd = "bgzip " + myvcf
whichcmd(cmd, args, 0)
# move extension from .gz to .bgz
cmd = "mv " + myvcf + ".gz " + myvcf + ".bgz "
whichcmd(cmd, args, 0)
cmd = "tabix -p vcf " + myvcf + ".bgz"
whichcmd(cmd, args, 0)
def run_vlad_code_compare(args, savidir, inplist):
"""Run Vladimir's savi computations to compare samples"""
# declare as global or else Python will treat as a local variable
global qvtargs
# variable to hold comma-delimited list of SGE job IDs
sgejobids = ""
# if hybrid, SAMP 1 tot depth = RD + AD, SAMP 2 tot depth = SDP
if (args.hybrid): qvtargs = "--hybrid"
# make sure the header file doesnt exist because we're going to append to it
cmd = "rm -f " + savidir + "/{header_addition.txt,vcfheader.txt}"
whichcmd(cmd, args, 0)
if (args.verbose): print("\n# savi comparison")
pairwise_list = [s for s in args.sample.split(",") if ":" in s]
for s in pairwise_list:
# the priors
str_a = s.split(":")[0]
str_b = s.split(":")[1]
prior_a = prior_dict[str_a]
prior_b = prior_dict[str_b]
# paste testout/sample_1_1.1p testout/sample_2_1.1p | cut -f1-4,6-8 | ../savi/savi_poster -pd testout/prior testout/prior | ../savi/savi_conf -fc 1e-5 > testout/savi/pd_12.txt
# paste testout/sample_1_1.1p testout/sample_3_1.1p | cut -f1-4,6-8 | ../savi/savi_poster -pd testout/prior testout/prior | ../savi/savi_conf -fc 1e-5 > testout/savi/pd_13.txt
# paste testout/sample_1_1.1p testout/sample_4_1.1p | cut -f1-4,6-8 | ../savi/savi_poster -pd testout/prior testout/prior | ../savi/savi_conf -fc 1e-5 > testout/savi/pd_14.txt
# paste testout/sample_2_1.1p testout/sample_3_1.1p | cut -f1-4,6-8 | ../savi/savi_poster -pd testout/prior testout/prior | ../savi/savi_conf -fc 1e-5 > testout/savi/pd_23.txt
# paste testout/sample_2_1.1p testout/sample_4_1.1p | cut -f1-4,6-8 | ../savi/savi_poster -pd testout/prior testout/prior | ../savi/savi_conf -fc 1e-5 > testout/savi/pd_24.txt
# paste testout/sample_3_1.1p testout/sample_4_1.1p | cut -f1-4,6-8 | ../savi/savi_poster -pd testout/prior testout/prior | ../savi/savi_conf -fc 1e-5 > testout/savi/pd_34.txt
cmd = "zcat " + savidir + "/freqsavi.vcf.bgz | " + \
args.bin + "/make_qvt -1 -2s " + str_a + "," + str_b + " " + qvtargs + " | " + \
args.bin + "/savi_poster -pd " + prior_a + " " + prior_b + " | " + \
args.bin + "/savi_conf -fs " + args.saviconf + " " + args.saviprecision + " | awk -v samp1=" + str_a + " -v samp2=" + str_b + " '" + '{mystr="PD"samp1 samp2; print mystr"_F="$1";"mystr"_L="$5";"mystr"_U="$(NF-1)}' + "' > " + savidir + "/pd_" + str_a + str_b + ".txt"
# args.bin + "/savi_conf -fc " + args.saviconf + " | awk -v samp1=" + str_a + " -v samp2=" + str_b + " '" + '{mystr="PD"samp1 samp2; print mystr"_F="$1";"mystr"_L="$5";"mystr"_U="$(NF-1)}' + "' > " + savidir + "/pd_" + str_a + str_b + ".txt"
cmd = escape_special_char(cmd)
# whichcmd(cmd, args, wantreturn, wantqsub=0, jobname="myjob", holdstr="0", wantsync=0):
# run cmd and store SGE job id
myjobid = whichcmd(cmd, args, 1, args.sge,
"j_s" + str_a + "_s" + str_b + "_cmp")
if (args.sge):
print("Your job " + myjobid + " has been submitted")
sgejobids = myjobid + "," + sgejobids
# fix vcf header
with open(savidir + "/header_addition.txt", 'a') as f:
f.write(
"##INFO=<ID=PD" + str_a + str_b +
"_F,Number=1,Type=Integer,Description=\"Savi freq delta for sample "
+ str_a + " vs " + str_b + "\">\n")
f.write(
"##INFO=<ID=PD" + str_a + str_b +
"_L,Number=1,Type=Integer,Description=\"Savi freq delta lower bound for sample "
+ str_a + " vs " + str_b + "\">\n")
f.write(
"##INFO=<ID=PD" + str_a + str_b +
"_U,Number=1,Type=Integer,Description=\"Savi freq delta upper bound for sample "
+ str_a + " vs " + str_b + "\">\n")
# a there was at least one savi comparison and there are more than 2 samples, run a 1 vs ALL comparision per order of JiGuang
if (pairwise_list and len(inplist) > 2):
# arbitarily use the first prior
str_a = pairwise_list[0].split(":")[0]
str_b = pairwise_list[0].split(":")[0]
prior_a = prior_dict[str_a]
prior_b = prior_dict[str_b]
cmd = "zcat " + savidir + "/freqsavi.vcf.bgz | " + \
args.bin + "/make_qvt -1 -1vsall | " + \
args.bin + "/savi_poster -pd " + prior_a + " " + prior_b + " | " + \
args.bin + "/savi_conf -fs " + args.saviconf + " " + args.saviprecision + " | awk -v samp1=0 -v samp2=0" + " '" + '{mystr="PD"samp1 samp2; print mystr"_F="$1";"mystr"_L="$5";"mystr"_U="$(NF-1)}' + "' > " + savidir + "/pd_00.txt"
cmd = escape_special_char(cmd)
# whichcmd(cmd, args, wantreturn, wantqsub=0, jobname="myjob", holdstr="0", wantsync=0):
# run cmd and store SGE job id
myjobid = whichcmd(cmd, args, 1, args.sge,
"j_s" + str_a + "_s" + str_b + "_cmp")
if (args.sge):
print("Your job " + myjobid + " has been submitted")
sgejobids = myjobid + "," + sgejobids
# fix vcf header
with open(savidir + "/header_addition.txt", 'a') as f:
f.write(
"##INFO=<ID=PD00_F,Number=1,Type=Integer,Description=\"Savi freq delta for samples 1 vs all the others\">\n"
)
f.write(
"##INFO=<ID=PD00_L,Number=1,Type=Integer,Description=\"Savi freq delta lower bound for samples 1 vs all the others\">\n"
)
f.write(
"##INFO=<ID=PD00_U,Number=1,Type=Integer,Description=\"Savi freq delta upper bound for samples 1 vs all the others\">\n"
)
# a there was at least one savi comparison
if (pairwise_list):
if (args.verbose): print("\n# paste savi numbers into INFO field")
cmd = 'paste -d";" ' + savidir + "/pd_*.txt > " + savidir + "/pd.txt"
cmd = escape_special_char(cmd)
# whichcmd(cmd, args, wantreturn, wantqsub=0, jobname="myjob", holdstr="0", wantsync=0):
mysyncjob = whichcmd(cmd, args, 1, args.sge, "mysync", sgejobids, 1)
if (args.sge):
print("Your job " + mysyncjob + " has been submitted")
if (not args.debug):
if (args.verbose): print("\n# clean up")
cmd = "rm -f " + savidir + "/pd_*.txt"
whichcmd(cmd, args, 0)
cmd = "zcat " + savidir + "/freqsavi.vcf.bgz | " + \
args.bin + "/add_to_info -f " + savidir + "/pd.txt --header " + savidir + "/vcfheader.txt > " + savidir + "/tmp0.txt"
whichcmd(cmd, args, 0)
cmd = "cat " + savidir + "/header_addition.txt " + savidir + "/vcfheader.txt " + savidir + "/tmp0.txt > " + savidir + "/finalsavi.vcf"
whichcmd(cmd, args, 0)
if (args.verbose): print("\n# compress finalsavi.vcf")
compress_vcf(args, savidir + "/finalsavi.vcf")
if (not args.debug):
if (not args.keepfreqfile):
if (args.verbose): print("\n# clean up")
cmd = "rm -f " + savidir + "/{freqsavi.vcf.bgz,freqsavi.vcf.bgz.tbi}"
whichcmd(cmd, args, 0)
if (not args.debug):
if (args.verbose): print("\n# clean up")
cmd = "rm -f " + savidir + "/{header_addition.txt,vcfheader.txt,tmp0.txt,pd.txt,addsavi.vcf.bgz,addsavi.vcf.bgz.tbi,filtersavi.vcf.bgz,filtersavi.vcf.bgz.tbi}"
whichcmd(cmd, args, 0)
if (args.verbose): print("[END]")
def run_vlad_code_freq(args, savidir, inplist):
"""Run Vladimir's savi computations to get freq"""
# declare as global or else Python will treat as a local variable
global qvtargs
# variable to hold comma-delimited list of SGE job IDs
sgejobids = ""
# this variable is true only for the first iteration of the list
bool_first = 1
# make sure the header file doesnt exist because we're going to append to it
cmd = "rm -f " + savidir + "/{header_addition.txt,vcfheader.txt}"
whichcmd(cmd, args, 0)
# prepend this to cmd str:
firstcmd = ""
if (args.nofilter):
firstcmd = "zcat " + savidir + "/addsavi.vcf.bgz | "
else:
firstcmd = "zcat " + savidir + "/filtersavi.vcf.bgz | "
# make freq files
for i in inplist:
if bool_first:
# if hybrid, start with tot depth = RD + AD for first iteration, then turn off
if (args.hybrid): qvtargs = "--rdplusad"
bool_first = 0
else:
# if hybrid, start with tot depth = RD + AD for first iteration, then turn off
if (args.hybrid): qvtargs = ""
if (args.verbose): print("\n# compute freq for sample " + i)
cmd = firstcmd + args.bin + "/make_qvt -1 -s " + i + " " + qvtargs + " | " + \
args.bin + "/savi_poster -p " + prior_dict[i] + " | " + \
args.bin + "/savi_conf -fs " + args.saviconf + " " + args.saviprecision + " | awk -v samp=" + i + " '" + '{mystr="P"samp; print mystr"_F="$1";"mystr"_L="$5";"mystr"_U="$(NF-1)}' + "' > " + savidir + "/freq_" + i + ".txt"
# args.bin + "/savi_conf -fc " + args.saviconf + " | awk -v samp=" + i + " '" + '{mystr="P"samp; print mystr"_F="$1";"mystr"_L="$5";"mystr"_U="$(NF-1)}' + "' > " + savidir + "/freq_" + i + ".txt"
cmd = escape_special_char(cmd)
# whichcmd(cmd, args, wantreturn, wantqsub=0, jobname="myjob", holdstr="0", wantsync=0):
# run cmd and store SGE job id
myjobid = whichcmd(cmd, args, 1, args.sge, "j_s" + i + "_freq")
if (args.sge):
print("Your job " + myjobid + " has been submitted")
sgejobids = myjobid + "," + sgejobids
# fix vcf header
with open(savidir + "/header_addition.txt", 'a') as f:
f.write(
"##INFO=<ID=P" + i +
"_F,Number=1,Type=Integer,Description=\"Savi freq for sample "
+ i + "\">\n")
f.write(
"##INFO=<ID=P" + i +
"_L,Number=1,Type=Integer,Description=\"Savi freq lower bound for sample "
+ i + "\">\n")
f.write(
"##INFO=<ID=P" + i +
"_U,Number=1,Type=Integer,Description=\"Savi freq upper bound for sample "
+ i + "\">\n")
if (args.verbose): print("\n# paste savi freq numbers into INFO field")
cmd = 'paste -d";" ' + savidir + "/freq_*.txt > " + savidir + "/freq.txt"
cmd = escape_special_char(cmd)
# whichcmd(cmd, args, wantreturn, wantqsub=0, jobname="myjob", holdstr="0", wantsync=0):
mysyncjob = whichcmd(cmd, args, 1, args.sge, "mysync", sgejobids, 1)
if (args.sge):
print("Your job " + mysyncjob + " has been submitted")
if (not args.debug):
if (args.verbose): print("\n# clean up")
cmd = "rm -f " + savidir + "/freq_*.txt"
whichcmd(cmd, args, 0)
# add into INFO field
cmd = firstcmd + args.bin + "/add_to_info -f " + savidir + "/freq.txt --header " + savidir + "/vcfheader.txt > " + savidir + "/tmp0.txt"
whichcmd(cmd, args, 0)
# add new header lines to vcf
cmd = "cat " + savidir + "/header_addition.txt " + savidir + "/vcfheader.txt " + savidir + "/tmp0.txt > " + savidir + "/freqsavi.vcf"
whichcmd(cmd, args, 0)
if (not args.debug):
if (args.verbose): print("\n# clean up")
cmd = "rm -f " + savidir + "/{header_addition.txt,vcfheader.txt,tmp0.txt,freq.txt}"
whichcmd(cmd, args, 0)
# compress
if (args.verbose): print("\n# compress freqsavi.vcf")
compress_vcf(args, savidir + "/freqsavi.vcf")