def blacklist_filter(peak, blacklist, keep_irregular_chr, out_dir):
prefix = os.path.join(out_dir, os.path.basename(strip_ext(peak)))
peak_ext = get_ext(peak)
filtered = '{}.bfilt.{}.gz'.format(prefix, peak_ext)
if get_num_lines(peak) == 0 or blacklist == '' \
or get_num_lines(blacklist) == 0:
cmd = 'zcat -f {} | gzip -nc > {}'.format(peak, filtered)
run_shell_cmd(cmd)
else:
# due to bedtools bug when .gz is given for -a and -b
tmp1 = gunzip(peak, 'tmp1', out_dir)
tmp2 = gunzip(blacklist, 'tmp2', out_dir)
cmd = 'bedtools intersect -nonamecheck -v -a {} -b {} | '
cmd += 'awk \'BEGIN{{OFS="\\t"}} '
cmd += '{{if ($5>1000) $5=1000; print $0}}\' | '
if not keep_irregular_chr:
cmd += 'grep -P \'chr[\\dXY]+\\b\' | '
cmd += 'gzip -nc > {}'
cmd = cmd.format(
tmp1, # peak
tmp2, # blacklist
filtered)
run_shell_cmd(cmd)
rm_f([tmp1, tmp2])
return filtered
def naive_overlap(basename_prefix, peak1, peak2, peak_pooled, peak_type,
nonamecheck, out_dir):
prefix = os.path.join(out_dir, basename_prefix)
prefix += '.overlap'
overlap_peak = '{}.{}.gz'.format(prefix, peak_type)
nonamecheck_param = '-nonamecheck' if nonamecheck else ''
if peak_type.lower() in ('narrowpeak', 'regionpeak'):
awk_param = '{s1=$3-$2; s2=$13-$12; if (($21/s1 >= 0.5) || ($21/s2 >= 0.5)) {print $0}}'
cut_param = '1-10'
elif peak_type.lower() == 'broadpeak':
awk_param = '{s1=$3-$2; s2=$12-$11; if (($19/s1 >= 0.5) || ($19/s2 >= 0.5)) {print $0}}'
cut_param = '1-9'
elif peak_type.lower() == 'gappedpeak':
awk_param = '{s1=$3-$2; s2=$18-$17; if (($31/s1 >= 0.5) || ($31/s2 >= 0.5)) {print $0}}'
cut_param = '1-15'
else:
raise ValueError('Unsupported peak_type.')
# due to bedtools bug when .gz is given for -a and -b
tmp1 = gunzip(peak1, 'tmp1', out_dir)
tmp2 = gunzip(peak2, 'tmp2', out_dir)
tmp_pooled = gunzip(peak_pooled, 'tmp_pooled', out_dir)
# Find pooled peaks that overlap peak1 and peak2
# where overlap is defined as the fractional overlap
# wrt any one of the overlapping peak pairs >= 0.5
cmd1 = 'intersectBed {} -wo '
cmd1 += '-a {} -b {} | '
cmd1 += 'awk \'BEGIN{{FS="\\t";OFS="\\t"}} {}\' | '
cmd1 += 'cut -f {} | sort | uniq | '
cmd1 += 'intersectBed {} -wo '
cmd1 += '-a stdin -b {} | '
cmd1 += 'awk \'BEGIN{{FS="\\t";OFS="\\t"}} {}\' | '
cmd1 += 'cut -f {} | sort | uniq | gzip -nc > {}'
cmd1 = cmd1.format(
nonamecheck_param,
tmp_pooled, # peak_pooled
tmp1, # peak1
awk_param,
cut_param,
nonamecheck_param,
tmp2, # peak2
awk_param,
cut_param,
overlap_peak)
run_shell_cmd(cmd1)
rm_f([tmp1, tmp2, tmp_pooled])
return overlap_peak
def frip(ta, peak, out_dir):
prefix = os.path.join(out_dir, os.path.basename(strip_ext(peak)))
frip_qc = '{}.frip.qc'.format(prefix)
if get_num_lines(peak) == 0:
val1 = 0.0
tmp_files = []
else:
# due to bedtools bug when .gz is given for -a and -b
tmp1 = gunzip(ta, 'tmp1', out_dir)
tmp2 = gunzip(peak, 'tmp2', out_dir)
cmd = 'bedtools intersect -nonamecheck -a {} -b {} -wa -u | wc -l'
cmd = cmd.format(
tmp1, # ta
tmp2) # peak
val1 = run_shell_cmd(cmd)
tmp_files = [tmp1, tmp2]
val2 = get_num_lines(ta)
write_txt(frip_qc, str(float(val1) / float(val2)))
rm_f(tmp_files)
return frip_qc
def blacklist_filter_bam(bam, blacklist, out_dir):
prefix = os.path.join(out_dir, os.path.basename(strip_ext_bam(bam)))
filtered = '{}.bfilt.bam'.format(prefix)
if blacklist == '' or get_num_lines(blacklist) == 0:
cmd = 'zcat -f {} | gzip -nc > {}'.format(bam, filtered)
run_shell_cmd(cmd)
else:
# due to bedtools bug when .gz is given for -a and -b
tmp2 = gunzip(blacklist, 'tmp2', out_dir)
cmd = 'bedtools intersect -nonamecheck -v -abam {} -b {} > {}'
cmd = cmd.format(
bam,
tmp2, # blacklist
filtered)
run_shell_cmd(cmd)
rm_f([tmp2])
return filtered
def frip_shifted(ta, peak, chrsz, fraglen, out_dir):
prefix = os.path.join(out_dir, os.path.basename(strip_ext(peak)))
frip_qc = '{}.frip.qc'.format(prefix)
half_fraglen = (fraglen + 1) / 2
if get_num_lines(peak) == 0:
val1 = 0.0
else:
# due to bedtools bug when .gz is given for -a and -b
tmp2 = gunzip(peak, 'tmp2', out_dir)
cmd = 'bedtools slop -i {} -g {} '
cmd += '-s -l {} -r {} | '
cmd += 'awk \'{{if ($2>=0 && $3>=0 && $2<=$3) print $0}}\' | '
cmd += 'bedtools intersect -nonamecheck -a stdin -b {} '
cmd += '-wa -u | wc -l'
cmd = cmd.format(ta, chrsz, -half_fraglen, half_fraglen, tmp2) # peak
val1 = run_shell_cmd(cmd)
rm_f(tmp2)
val2 = get_num_lines(ta)
write_txt(frip_qc, str(float(val1) / float(val2)))
return frip_qc