def vcf_to_h5(vcf_fpath, out_h5_fpath, n_threads, preread_nvars, tmp_dir,
kept_fields=None, ignored_fields=None):
if not os.path.exists(tmp_dir):
os.mkdir(tmp_dir)
chroms = get_chroms_in_vcf(vcf_fpath)
max_field_lens, max_field_str_lens = _get_max_field(vcf_fpath,
preread_nvars,
kept_fields=kept_fields,
ignored_fields=ignored_fields)
partial_parse_vcf = partial(_parse_vcf, vcf_fpath=vcf_fpath,
tmp_dir=tmp_dir,
max_field_lens=max_field_lens,
max_field_str_lens=max_field_str_lens,
kept_fields=kept_fields,
ignored_fields=ignored_fields)
with Pool(n_threads) as pool:
try:
h5_chroms_fpaths = pool.map(partial_parse_vcf, chroms)
except Exception:
remove_temp_file_in_dir(tmp_dir, '.tmp.h5')
raise
try:
_merge_h5(h5_chroms_fpaths, out_h5_fpath)
except Exception:
raise
finally:
_remove_temp_chrom_h5s(h5_chroms_fpaths)
def write_vcf_parallel(variations, out_fhand, n_threads, tmp_dir,
chunk_size=None, vcf_format=VCF_FORMAT):
_write_vcf_meta(variations, out_fhand, vcf_format=vcf_format)
_write_vcf_header(variations, out_fhand)
grouped_paths = _group_variations_paths(variations)
_partial_write_snvs = partial(_write_snvs_parallel, tmp_dir=tmp_dir,
grouped_paths=grouped_paths)
with Pool(n_threads) as pool:
try:
vcf_fpaths = pool.map(_partial_write_snvs,
_numbered_chunks(variations, chunk_size))
except Exception:
remove_temp_file_in_dir(tmp_dir, '.vcf.h5')
raise
try:
_merge_vcfs(vcf_fpaths, out_fhand)
except Exception:
raise
finally:
for vcf_fpath in vcf_fpaths:
if os.path.exists(vcf_fpath):
os.remove(vcf_fpath)
def write_vcf_parallel(variations, out_fhand, n_threads, tmp_dir,
chunk_size=None, vcf_format=VCF_FORMAT):
_write_vcf_meta(variations, out_fhand, vcf_format=vcf_format)
_write_vcf_header(variations, out_fhand)
grouped_paths = _group_variations_paths(variations)
_partial_write_snvs = partial(_write_snvs_parallel, tmp_dir=tmp_dir,
grouped_paths=grouped_paths)
with Pool(n_threads) as pool:
try:
vcf_fpaths = pool.map(_partial_write_snvs,
_numbered_chunks(variations, chunk_size))
except Exception:
remove_temp_file_in_dir(tmp_dir, '.vcf.h5')
raise
try:
_merge_vcfs(vcf_fpaths, out_fhand)
except Exception:
raise
finally:
for vcf_fpath in vcf_fpaths:
if os.path.exists(vcf_fpath):
os.remove(vcf_fpath)
def vcf_to_h5(vcf_fpath, out_h5_fpath, n_threads, tmp_dir, kept_fields=None,
ignored_fields=None):
if not os.path.exists(tmp_dir):
os.mkdir(tmp_dir)
chroms = get_chroms_in_vcf(vcf_fpath)
partial_parse_vcf = partial(_parse_vcf, vcf_fpath=vcf_fpath,
tmp_dir=tmp_dir,
kept_fields=kept_fields,
ignored_fields=ignored_fields)
with Pool(n_threads) as pool:
try:
h5_chroms_fpaths = pool.map(partial_parse_vcf, chroms)
except Exception:
remove_temp_file_in_dir(tmp_dir, '.tmp.h5')
raise
try:
_merge_h5(h5_chroms_fpaths, out_h5_fpath)
except Exception:
raise
finally:
_remove_temp_chrom_h5s(h5_chroms_fpaths)
