def write_all_wiggles(files1, files2, corrected_group, basecall_subgroups,
obs_filter, test_type, min_test_vals, stats_fn,
fishers_method_offset, wig_base, wig_types):
stats_file_exists = stats_fn is not None and os.path.isfile(stats_fn)
include_stats = 'pvals' in wig_types or 'qvals' in wig_types
if include_stats and stats_file_exists:
if VERBOSE: sys.stderr.write('Loading statistics from file.\n')
all_stats = ns.parse_stats(stats_fn)
if VERBOSE: sys.stderr.write('Parsing FAST5 files.\n')
raw_read_coverage1 = nh.parse_fast5s(files1, corrected_group,
basecall_subgroups)
raw_read_coverage1 = nh.filter_reads(raw_read_coverage1, obs_filter)
group1_name = '' if files2 is None else 'group1'
if files2 is not None:
raw_read_coverage2 = nh.parse_fast5s(files2, corrected_group,
basecall_subgroups)
raw_read_coverage2 = nh.filter_reads(raw_read_coverage2, obs_filter)
if include_stats and not stats_file_exists:
if VERBOSE: sys.stderr.write('Calculating statistics.\n')
all_stats = ns.get_all_significance(raw_read_coverage1,
raw_read_coverage2, test_type,
min_test_vals, stats_fn,
fishers_method_offset)
if VERBOSE: sys.stderr.write('Writing wiggles.\n')
if 'coverage' in wig_types:
write_cov_wig(raw_read_coverage2, wig_base, GROUP2_NAME)
if 'signal_sd' in wig_types:
write_signal_sd_wig(raw_read_coverage2, wig_base, GROUP2_NAME)
if 'length' in wig_types:
write_length_wig(raw_read_coverage2, wig_base, GROUP2_NAME)
# need to do signal and difference call once either with or
# w/o second set of files (unlike coverage, sds and length
if 'signal' in wig_types or 'difference' in wig_types:
write_signal_and_diff_wigs(raw_read_coverage1, raw_read_coverage2,
wig_base, group1_name, 'signal'
in wig_types, 'difference' in wig_types)
else:
if VERBOSE: sys.stderr.write('Writing wiggles.\n')
if 'signal' in wig_types or 'difference' in wig_types:
write_signal_and_diff_wigs(raw_read_coverage1, None, wig_base,
group1_name, 'signal' in wig_types,
'difference' in wig_types)
if 'coverage' in wig_types:
write_cov_wig(raw_read_coverage1, wig_base, group1_name)
if 'signal_sd' in wig_types:
write_signal_sd_wig(raw_read_coverage1, wig_base, group1_name)
if 'length' in wig_types:
write_length_wig(raw_read_coverage1, wig_base, group1_name)
if 'pvals' in wig_types or 'qvals' in wig_types:
write_pvals_and_qvals_wig(all_stats, wig_base, 'pvals' in wig_types,
'qvals' in wig_types)
return
def write_most_signif(files1, files2, num_regions, qval_thresh,
corrected_group, basecall_subgroups, seqs_fn, num_bases,
test_type, obs_filter, min_test_vals, stats_fn, fasta_fn,
fishers_method_offset):
calc_stats = stats_fn is None or not os.path.isfile(stats_fn)
if not calc_stats:
if VERBOSE: sys.stderr.write('Loading statistics from file.\n')
all_stats = ns.parse_stats(stats_fn)
if calc_stats or fasta_fn is None:
if VERBOSE: sys.stderr.write('Parsing files.\n')
raw_read_coverage1 = nh.parse_fast5s(files1, corrected_group,
basecall_subgroups)
raw_read_coverage2 = nh.parse_fast5s(files2, corrected_group,
basecall_subgroups)
raw_read_coverage1 = nh.filter_reads(raw_read_coverage1, obs_filter)
raw_read_coverage2 = nh.filter_reads(raw_read_coverage2, obs_filter)
if calc_stats:
if VERBOSE: sys.stderr.write('Calculating statistics.\n')
all_stats = ns.get_all_significance(raw_read_coverage1,
raw_read_coverage2, test_type,
min_test_vals, stats_fn,
fishers_method_offset)
plot_intervals = ns.get_most_signif_regions(all_stats, num_bases,
num_regions, qval_thresh)
if fasta_fn is None:
reg_seqs = get_region_sequences(plot_intervals, raw_read_coverage1,
raw_read_coverage2, num_bases,
corrected_group)
else:
fasta_records = nh.parse_fasta(fasta_fn)
reg_seqs = [(p_int, fasta_records[chrm][start:start + num_bases])
for p_int, (chrm, start, strand,
reg_name) in plot_intervals
if chrm in fasta_records]
# get reads overlapping each region
if VERBOSE: sys.stderr.write('Outputting region seqeuences.\n')
with open(seqs_fn, 'w') as seqs_fp:
for reg_i, reg_seq in reg_seqs:
chrm, start, strand, stat = next(
p_int for p_reg_i, p_int in plot_intervals if p_reg_i == reg_i)
if strand == '-':
reg_seq = nh.rev_comp(reg_seq)
seqs_fp.write('>{0}::{1:d}::{2} {3}\n{4}\n'.format(
chrm, start, strand, stat, ''.join(reg_seq)))
return
def write_wiggle(files, corrected_group, wiggle_fn, basecall_subgroups,
obs_filter):
if VERBOSE: sys.stderr.write('Parsing files.\n')
raw_read_coverage = parse_fast5s(files, corrected_group,
basecall_subgroups)
raw_read_coverage = filter_reads(raw_read_coverage, obs_filter)
if VERBOSE: sys.stderr.write('Calculating read coverage.\n')
wiggle_cov = []
for chrom, reads_data in raw_read_coverage.items():
max_end = max(r_data.end for r_data in reads_data)
chrom_coverage = np.zeros(max_end, dtype=np.int_)
for r_data in reads_data:
chrom_coverage[r_data.start:r_data.end] += 1
wiggle_cov.append((chrom, chrom_coverage))
if VERBOSE: sys.stderr.write('Writing wiggle.\n')
with open(wiggle_fn, 'w') as wig_fp:
wig_fp.write(
'track type=wiggle_0 name={0} description={0}\n'.format(wiggle_fn))
for chrm, chrm_cov in wiggle_cov:
wig_fp.write("variableStep chrom={} span=1\n".format(chrm))
wig_fp.write('\n'.join([
str(int(pos) + 1) + " " + str(int(val))
for pos, val in enumerate(chrm_cov) if val > 0
]) + '\n')
return