def main(chrs, scale, step, meth_dir, cpg_dir, out_dir, sname):
LOGFH = open('create_microarray_tracks.LOG', 'w')
LOGFH.write('Start the program ... %s\n' % datetime.datetime.now())
# Create methylation tracks for each chromosome
for chr in chrs:
# 1. Build CpGs
LOGFH.write('... Build CpG sites for [ %s ] ... %s\n' % (chr, datetime.datetime.now()))
try:
cpgfile = check_file(chr+'_cpgs', PATH=cpg_dir)
except MethError, e:
print >> sys.stderr, e.value
continue
siteparser = SiteParser(chr)
siteparser.parse_sites(cpgfile, 'CpGSimple')
cpg_sites = siteparser.get_sites()
for cpg in cpg_sites: # define empty meth score
cpg.meth_score = '-10000'
# 2. Get methylation scores
# note files are saved in 'methstatus_chrN'
LOGFH.write('... Obtain methylation scores ...\n')
try:
mfile = check_file('methstatus_' + chr, PATH=meth_dir)
except MethError, e:
print >> sys.stderr, e.value
sys.exit(2)
def main(chrs, scale, step, meth_dir, cpg_dir, out_dir, sname):
LOGFH = open('create_microarray_tracks.LOG', 'w')
LOGFH.write('Start the program ... %s\n' % datetime.datetime.now())
# Create methylation tracks for each chromosome
for chr in chrs:
# 1. Build CpGs
LOGFH.write('... Build CpG sites for [ %s ] ... %s\n' %
(chr, datetime.datetime.now()))
try:
cpgfile = check_file(chr + '_cpgs', PATH=cpg_dir)
except MethError, e:
print >> sys.stderr, e.value
continue
siteparser = SiteParser(chr)
siteparser.parse_sites(cpgfile, 'CpGSimple')
cpg_sites = siteparser.get_sites()
for cpg in cpg_sites: # define empty meth score
cpg.meth_score = '-10000'
# 2. Get methylation scores
# note files are saved in 'methstatus_chrN'
LOGFH.write('... Obtain methylation scores ...\n')
try:
mfile = check_file('methstatus_' + chr, PATH=meth_dir)
except MethError, e:
print >> sys.stderr, e.value
sys.exit(2)
def main(chr, seqfile, outname):
# Initiate the SiteParser
LOGFH = open('parse_sites.LOG', 'w')
LOGFH.write('Start the program for [%s] ... %s\n' % (chr, str(datetime.datetime.now())))
LOGFH.flush()
siteparser = SiteParser(chr)
# Parse sites
LOGFH.write('... Start to parse sites ... %s\n' % str(datetime.datetime.now()))
LOGFH.flush()
siteparser.parse_seq(seqfile)
# Write to the output files
LOGFH.write('... Save annotation files ... %s\n' % str(datetime.datetime.now()))
LOGFH.flush()
# 1. CpGs
write2output(outname+'_cpgs', siteparser.sites, chr)
# 2. RE
write2output(outname+'_re', siteparser.re_sites, chr)
# 3. McrBC
write2output(outname+'_mcrbc', siteparser.mcrbc_sites, chr)
LOGFH.write('Finish the program for [%s] ... %s\n' % (chr, str(datetime.datetime.now())))
LOGFH.close()
# Get annotation files: e.g., chr1_cpgs, chr1_re, chr1_mcrbc
try:
cpg_file = check_file(chr+'_cpgs', PATH=anno_dir)
re_sitefile = check_file(chr+'_re', PATH=anno_dir)
mcrbc_sitefile = check_file(chr+'_mcrbc', PATH=anno_dir)
except MethError, e:
print >> sys.stderr, e.value
sys.exit(2)
# 1. Read parameters
LOGFH = open(os.path.join(out_dir, 'filter.LOG'), 'a')
LOGFH.write('Start the program for [%s] ... %s\n' % (chr, str(datetime.datetime.now())))
LOGFH.flush()
paras = parse_paras(parafile)
# 2. Build CpGs
LOGFH.write('... Build CpG sites ... %s\n' % str(datetime.datetime.now()))
siteparser = SiteParser(chr)
siteparser.parse_sites(cpg_file, 'CpGFull')
cpg_sites = siteparser.get_sites()
# 3. Build RE sites first
LOGFH.write('... Build RE sites ... %s\n' % str(datetime.datetime.now()))
siteparser = SiteParser(chr)
siteparser.parse_sites(re_sitefile, 'RE')
re_sites = siteparser.get_sites()
del siteparser
# 4. Build RE fragments
LOGFH.write('... Build RE fragments ... %s\n' % str(datetime.datetime.now()))
fragparser = FragmentParser()
# Split fragfile into subsets to avoid memory overflow
re_linenum = count_file_lines(re_fragfile)
refh = open(re_fragfile)
re_fraglines = []