def get_parsed_files_list(d):
fl = fn.get_files(d)
pfl = parse_files_list(fl)
return pfl
inp1 = directory with maf and mod files
inp2 = Newick tree file
inp3 = output stump
'''
import os, sys, fn
dir = sys.argv[1]
newick = os.path.abspath(sys.argv[2])
out_nm = sys.argv[3]
def copy_tree__write_runccs(tree, fl, m_runcc, p_runcc):
new_tree = "%s.%s.nwk" % (tree, fl.split("/")[-1].replace(".maf", ""))
copy_command = "cp %s %s" % (tree, new_tree)
os.system(copy_command)
model_command = "module load PHAST;phastCons -i MAF --estimate-trees %s --no-post-probs %s %s.mod\n"%\
(new_tree,fl,fl)
m_runcc.write(model_command)
phastCons_command = "module load PHAST;phastCons -i MAF %s %s.cons.mod,%s.noncons.mod > %s.wig\n"%\
(fl,new_tree,new_tree,fl)
p_runcc.write(phastCons_command)
files_list = fn.get_files(dir, ".maf", "tail")
model_runcc = open(out_nm + "-generate_models.runcc", "w")
phastCons_runcc = open(out_nm + "-run_phastCons.runcc", "w")
for file in files_list:
copy_tree__write_runccs(newick, file, model_runcc, phastCons_runcc)
model_runcc.close()
phastCons_runcc.close()
print '''
inp1 = directory with .lav files
inp2 = output directory for .psl files
'''
import os, sys, fn
dir = sys.argv[1]
out_dir = os.path.abspath(sys.argv[2])
if os.path.isfile("lav2psl.sh") == True:
chain_fl = fn.get_files(".", "chain_psl")
iteration = len(chain_fl) + 1
qsub_name = "lav2psl_%i.sh" % iteration
shell = open(qsub_name, "w")
else:
qsub_name = "lav2psl.sh"
shell = open(qsub_name, "w")
shell.write('''#!/bin/bash -login
#PBS -q main
#PBS -l nodes=1:ppn=1,walltime=02:00:00,mem=1gb
''')
files_list = fn.get_files(dir, ".lav", "tail")
for file in files_list:
command_line = "/mnt/home/lloydjo1/bin/liftOver/lavToPsl %s %s/%s\n" % (
file, out_dir, file.split("/")[-1].replace("lav", "psl"))
# os.system(command_line)
shell.write(command_line)
shell.close()
os.system("qsub %s" % qsub_name)
print '''
inp1 = directory with .chain files
inp2 = output name
'''
import os, sys, fn
dir = os.path.abspath(sys.argv[1])
out_name = dir + "/" + sys.argv[2]
if not out_name.endswith(".chain.sorted"):
out_name = out_name + ".chain.sorted"
if os.path.isfile("chainMergeSort.sh") == True:
chain_fl = fn.get_files(".", "chainMergeSort")
iteration = len(chain_fl) + 1
qsub_name = "chainMergeSort_%i.sh" % iteration
shell = open(qsub_name, "w")
else:
qsub_name = "chainMergeSort.sh"
shell = open(qsub_name, "w")
shell.write('''#!/bin/bash -login
#PBS -q main
#PBS -l nodes=1:ppn=1,walltime=01:00:00,mem=1gb
/mnt/home/lloydjo1/bin/liftOver/chainMergeSort %s/*.chain.sorted.merged > %s
''' % (dir, out_name))
shell.close()
os.system("qsub %s" % qsub_name)
print '''
inp1 = directory with fasta files
'''
import os, sys, fn
dir = sys.argv[1]
def parse_files_list(fl):
l = []
for item in fl:
if item.endswith("lav") or item.endswith("psl") or item.endswith(
"nib") or item.endswith(".sizes"):
pass
else:
l.append(item)
return l
files_list = fn.get_files(dir)
pfl = parse_files_list(files_list)
for file in pfl:
print "Working on file:", file
command_line = "/mnt/home/lloydjo1/bin/liftOver/faSize %s -detailed > %s.sizes" % (
file, file)
os.system(command_line)
print'''
inp1 = maf file 1
inp2 = maf file 2
inp3 = output name
'''
import os,sys,fn
maf1 = os.path.abspath(sys.argv[1])
maf2 = os.path.abspath(sys.argv[2])
out_nm = os.path.abspath(sys.argv[3])
if os.path.isfile("multiz_submit.sh") == True:
chain_fl = fn.get_files(".","multiz_submit")
iteration = len(chain_fl)+1
qsub_name = "multiz_submit%i.sh"%iteration
shell = open(qsub_name,"w")
else:
qsub_name = "multiz_submit.sh"
shell = open(qsub_name,"w")
shell.write('''#!/bin/bash -login
#PBS -q main
#PBS -l nodes=1:ppn=1,walltime=01:00:00,mem=1gb
/mnt/home/lloydjo1/bin/multiz-tba.012109/multiz %s %s v > %s
'''%(maf1,maf2,out_nm))
shell.close()
os.system("qsub %s"%qsub_name)