def bin_fast5_file(f5_path, tax_annot_res_dir, sens, min_qual, min_qlen,
min_pident, min_coverage, no_trash):
# Function bins FAST5 file with untwisting.
#
# :param f5_path: path to FAST5 file meant to be processed;
# :type f5_path: str;
# :param tax_annot_res_dir: path to directory containing taxonomic annotation;
# :type tax_annot_res_dir: str;
# :param sens: binning sensitivity;
# :type sens: str;
# :param min_qual: threshold for quality filter;
# :type min_qual: float;
# :param min_qlen: threshold for length filter;
# :type min_qlen: int (or None, if this filter is disabled);
# :param min_pident: threshold for alignment identity filter;
# :type min_pident: float (or None, if this filter is disabled);
# :param min_coverage: threshold for alignment coverage filter;
# :type min_coverage: float (or None, if this filter is disabled);
# :param no_trash: loical value. True if user does NOT want to output trash files;
# :type no_trash: bool;
outdir_path = os.path.dirname(
logging.getLoggerClass().root.handlers[0].baseFilename)
seqs_pass = 0 # counter for sequences, which pass filters
QL_seqs_fail = 0 # counter for too short or too low-quality sequences
align_seqs_fail = 0 # counter for sequences, which align to their best hit with too low identity or coverage
srt_file_dict = dict()
index_dirpath = os.path.join(
tax_annot_res_dir,
index_name) # name of directory that will contain indicies
# Make filter for quality and length
QL_filter = get_QL_filter(f5_path, min_qual, min_qlen)
# Configure path to trash file
if not no_trash:
QL_trash_fpath = get_QL_trash_fpath(
f5_path,
outdir_path,
min_qual,
min_qlen,
)
else:
QL_trash_fpath = None
# end if
# Make filter for identity and coverage
align_filter = get_align_filter(min_pident, min_coverage)
# Configure path to this trash file
if not no_trash:
align_trash_fpath = get_align_trash_fpath(f5_path, outdir_path,
min_pident, min_coverage)
else:
align_trash_fpath = None
# end if
# File validation:
# RuntimeError will be raised if FAST5 file is broken.
try:
# File existance checking is performed while parsing CL arguments.
# Therefore, this if-statement will trigger only if f5_path's file is not a valid HDF5 file.
if not h5py.is_hdf5(f5_path):
raise RuntimeError("file is not of HDF5 (i.e. not FAST5) format")
# end if
from_f5 = h5py.File(f5_path, 'r')
for _ in from_f5:
break
# end for
except RuntimeError as runterr:
printlog_error_time("FAST5 file is broken")
printlog_error("Reading the file `{}` crashed.".format(
os.path.basename(f5_path)))
printlog_error("Reason: {}".format(str(runterr)))
printlog_error("Omitting this file...")
print()
# Return zeroes -- inc_val won't be incremented and this file will be omitted
return (0, 0, 0)
# end try
# singleFAST5 and multiFAST5 files should be processed in different ways
# "Raw" group always in singleFAST5 root and never in multiFAST5 root
if "Raw" in from_f5.keys():
f5_cpy_func = copy_single_f5
else:
f5_cpy_func = copy_read_f5_2_f5
# end if
readids_to_seek = list(from_f5.keys()) # list of not-binned-yet read IDs
# Fill the list 'readids_to_seek'
for read_name in fast5_readids(from_f5):
# Get rid of "read_"
readids_to_seek.append(sys.intern(read_name))
# end for
# Walk through the index
index_f5_2_tsv = open_shelve(os.path.join(index_dirpath, index_name), 'r')
if not f5_path in index_f5_2_tsv.keys():
printlog_error_time(
"Source FAST5 file `{}` not found in index".format(f5_path))
printlog_error("Try to rebuild index")
platf_depend_exit(1)
# end if
for tsv_path in index_f5_2_tsv[f5_path].keys():
read_names = index_f5_2_tsv[f5_path][tsv_path]
taxonomy_path = os.path.join(tax_annot_res_dir, "taxonomy",
"taxonomy.tsv")
resfile_lines = configure_resfile_lines(tsv_path, sens, taxonomy_path)
for read_name in read_names:
try:
hit_names, *vals_to_filter = resfile_lines[sys.intern(
fmt_read_id(read_name)[1:])]
except KeyError:
printlog_error_time("Error: missing taxonomic annotation info for read `{}`"\
.format(fmt_read_id(read_name)[1:]))
printlog_error(
"It is stored in `{}` FAST5 file".format(f5_path))
printlog_error(
"Try to make new index file (press ENTER on corresponding prompt)."
)
printlog_error(
"Or, if does not work for you, make sure that taxonomic annotation info \
for this read is present in one of TSV files generated by `barapost-prober.py` and `barapost-local.py`."
)
index_f5_2_tsv.close()
platf_depend_exit(1)
# end try
if not QL_filter(vals_to_filter):
# Get name of result FASTQ file to write this read in
if QL_trash_fpath not in srt_file_dict.keys():
srt_file_dict = update_file_dict(srt_file_dict,
QL_trash_fpath)
# end if
f5_cpy_func(from_f5, read_name, srt_file_dict[QL_trash_fpath])
QL_seqs_fail += 1
elif not align_filter(vals_to_filter):
# Get name of result FASTQ file to write this read in
if align_trash_fpath not in srt_file_dict.keys():
srt_file_dict = update_file_dict(srt_file_dict,
align_trash_fpath)
# end if
f5_cpy_func(from_f5, read_name,
srt_file_dict[align_trash_fpath])
align_seqs_fail += 1
else:
for hit_name in hit_names.split(
"&&"
): # there can be multiple hits for single query sequence
# Get name of result FASTQ file to write this read in
binned_file_path = os.path.join(
outdir_path, "{}.fast5".format(hit_name))
if binned_file_path not in srt_file_dict.keys():
srt_file_dict = update_file_dict(
srt_file_dict, binned_file_path)
# end if
f5_cpy_func(from_f5, read_name,
srt_file_dict[binned_file_path])
# end for
seqs_pass += 1
# end if
# end for
from_f5.close()
index_f5_2_tsv.close()
# Close all binned files
for file_obj in filter(lambda x: not x is None, srt_file_dict.values()):
file_obj.close()
# end for
sys.stdout.write('\r')
printlog_info_time("File `{}` is binned.".format(
os.path.basename(f5_path)))
printn(" Working...")
return (seqs_pass, QL_seqs_fail, align_seqs_fail)
def map_f5reads_2_taxann(f5_path, tsv_taxann_lst, tax_annot_res_dir):
# Function perform mapping of all reads stored in input FAST5 files
# to existing TSV files containing taxonomic annotation info.
#
# It creates an DBM index file.
#
# :param f5_path: path to current FAST5 file;
# :type f5_path: str;
# :param tsv_taxann_lst: list of path to TSV files that contain taxonomic annotation;
# :type tsv_taxann_lst: list<str>;
# :param tax_annot_res_dir: path to directory containing taxonomic annotation;
# :type tax_annot_res_dir: str;
index_dirpath = os.path.join(
tax_annot_res_dir,
index_name) # name of directory that will contain indicies
# File validation:
# RuntimeError will be raised if FAST5 file is broken.
try:
# File existance checking is performed while parsing CL arguments.
# Therefore, this if-statement will trigger only if f5_path's file is not a valid HDF5 file.
if not h5py.is_hdf5(f5_path):
raise RuntimeError("file is not of HDF5 (i.e. not FAST5) format")
# end if
f5_file = h5py.File(f5_path, 'r')
for _ in f5_file:
break
# end for
except RuntimeError as runterr:
printlog_error_time("FAST5 file is broken")
printlog_error("Reading the file `{}` crashed.".format(
os.path.basename(f5_path)))
printlog_error("Reason: {}".format(str(runterr)))
printlog_error("Omitting this file...")
print()
return
# end try
readids_to_seek = list(fast5_readids(f5_file))
idx_dict = dict() # dictionary for index
# This saving is needed to compare with 'len(readids_to_seek)'
# after all TSV will be looked through in order to
# determine if some reads miss taxonomic annotation.
len_before = len(readids_to_seek)
# Iterate over TSV-taaxnn file
for tsv_taxann_fpath in tsv_taxann_lst:
with open(tsv_taxann_fpath, 'r') as taxann_file:
# Get all read IDs in current TSV
readids_in_tsv = list(
map(lambda l: l.split('\t')[0], taxann_file.readlines()))
# Iterate over all other reads in current FAST5
# ('reversed' is necessary because we remove items from list in this loop)
for readid in reversed(readids_to_seek):
fmt_id = fmt_read_id(readid)[1:]
if fmt_id in readids_in_tsv:
# If not first -- write data to dict (and to index later)
try:
idx_dict[tsv_taxann_fpath].append(
"read_" + fmt_id) # append to existing list
except KeyError:
idx_dict[tsv_taxann_fpath] = ["read_" + fmt_id
] # create a new list
finally:
readids_to_seek.remove(readid)
# end try
# end if
# end for
# end with
if len(readids_to_seek) == 0:
break
# end if
# end for
# If after all TSV is checked but nothing have changed -- we miss taxonomic annotation
# for some reads! And we will write their IDs to 'missing_reads_lst.txt' file.
if len(readids_to_seek) == len_before:
printlog_error_time("reads from FAST5 file not found")
printlog_error("FAST5 file: `{}`".format(f5_path))
printlog_error("Some reads have not undergone taxonomic annotation.")
missing_log = "missing_reads_lst.txt"
printlog_error("List of missing reads are in following file:")
printlog_error("{}".format(missing_log))
with open(missing_log, 'w') as missing_logfile:
missing_logfile.write(
"Missing reads from file '{}':\n\n".format(f5_path))
for readid in readids_to_seek:
missing_logfile.write(fmt_read_id(readid) + '\n')
# end for
try:
for path in glob(os.path.join(index_dirpath, '*')):
os.unlink(path)
# end for
os.rmdir(index_dirpath)
except OSError as oserr:
printlog_error(
"Error occured while removing index directory: {}".format(
oserr))
finally:
platf_depend_exit(3)
# end try
# end if
try:
# Open index files appending to existing data ('c' parameter)
with open_shelve(os.path.join(index_dirpath, index_name),
'c') as index_f5_2_tsv:
# Update index
index_f5_2_tsv[f5_path] = idx_dict
# end with
except OSError as oserr:
printlog_error_time("Error: cannot create index file `{}`"\
.format(os.path.join(index_dirpath, index_name)))
printlog_error(str(oserr))
platf_depend_exit(1)
def bin_fast5_file(f5_path, tax_annot_res_dir, sens, min_qual, min_qlen,
min_pident, min_coverage, no_trash):
# Function bins FAST5 file without untwisting.
#
# :param f5_path: path to FAST5 file meant to be processed;
# :type f5_path: str;
# :param tax_annot_res_dir: path to directory containing taxonomic annotation;
# :type tax_annot_res_dir: str;
# :param sens: binning sensitivity;
# :type sens: str;
# :param min_qual: threshold for quality filter;
# :type min_qual: float;
# :param min_qlen: threshold for length filter;
# :type min_qlen: int (or None, if this filter is disabled);
# :param min_pident: threshold for alignment identity filter;
# :type min_pident: float (or None, if this filter is disabled);
# :param min_coverage: threshold for alignment coverage filter;
# :type min_coverage: float (or None, if this filter is disabled);
# :param no_trash: loical value. True if user does NOT want to output trash files;
# :type no_trash: bool;
outdir_path = os.path.dirname(logging.getLoggerClass().root.handlers[0].baseFilename)
seqs_pass = 0 # counter for sequences, which pass filters
QL_seqs_fail = 0 # counter for too short or too low-quality sequences
align_seqs_fail = 0 # counter for sequences, which align to their best hit with too low identity or coverage
srt_file_dict = dict()
new_dpath = glob("{}{}*{}*".format(tax_annot_res_dir, os.sep, get_checkstr(f5_path)))[0]
tsv_res_fpath = get_res_tsv_fpath(new_dpath)
taxonomy_path = os.path.join(tax_annot_res_dir, "taxonomy", "taxonomy.tsv")
resfile_lines = configure_resfile_lines(tsv_res_fpath, sens, taxonomy_path)
# Make filter for quality and length
QL_filter = get_QL_filter(f5_path, min_qual, min_qlen)
# Configure path to trash file
if not no_trash:
QL_trash_fpath = get_QL_trash_fpath(f5_path, outdir_path, min_qual, min_qlen,)
else:
QL_trash_fpath = None
# end if
# Make filter for identity and coverage
align_filter = get_align_filter(min_pident, min_coverage)
# Configure path to this trash file
if not no_trash:
align_trash_fpath = get_align_trash_fpath(f5_path, outdir_path, min_pident, min_coverage)
else:
align_trash_fpath = None
# end if
# File validation:
# RuntimeError will be raised if FAST5 file is broken.
try:
# File existance checking is performed while parsing CL arguments.
# Therefore, this if-statement will trigger only if f5_path's file is not a valid HDF5 file.
if not h5py.is_hdf5(f5_path):
raise RuntimeError("file is not of HDF5 (i.e. not FAST5) format")
# end if
from_f5 = h5py.File(f5_path, 'r')
for _ in from_f5:
break
# end for
except RuntimeError as runterr:
printlog_error_time("FAST5 file is broken")
printlog_error("Reading the file `{}` crashed.".format(os.path.basename(f5_path)))
printlog_error("Reason: {}".format( str(runterr) ))
printlog_error("Omitting this file...")
print()
# Return zeroes -- inc_val won't be incremented and this file will be omitted
return (0, 0, 0)
# end try
# singleFAST5 and multiFAST5 files should be processed in different ways
# "Raw" group always in singleFAST5 root and never in multiFAST5 root
if "Raw" in from_f5.keys():
f5_cpy_func = copy_single_f5
else:
f5_cpy_func = copy_read_f5_2_f5
# end if
for _, read_name in enumerate(fast5_readids(from_f5)):
try:
hit_names, *vals_to_filter = resfile_lines[sys.intern(fmt_read_id(read_name))[1:]] # omit 'read_' in the beginning of FAST5 group's name
except KeyError:
printlog_error_time("Error: read `{}` not found in TSV file containing taxonomic annotation."\
.format(fmt_read_id(read_name)))
printlog_error("This TSV file: `{}`".format(tsv_res_fpath))
printlog_error("Try running barapost-binning with `-u` (`--untwist-fast5`) flag.\n")
platf_depend_exit(1)
# end try
# If read is found in TSV file:
if not QL_filter(vals_to_filter):
QL_seqs_fail += 1
# Get name of result FASTQ file to write this read in
if QL_trash_fpath not in srt_file_dict.keys():
srt_file_dict = update_file_dict(srt_file_dict, QL_trash_fpath)
# end if
f5_cpy_func(from_f5, read_name, srt_file_dict[QL_trash_fpath])
elif not align_filter(vals_to_filter):
align_seqs_fail += 1
# Get name of result FASTQ file to write this read in
if QL_trash_fpath not in srt_file_dict.keys():
srt_file_dict = update_file_dict(srt_file_dict, align_trash_fpath)
# end if
f5_cpy_func(from_f5, read_name, srt_file_dict[align_trash_fpath])
else:
for hit_name in hit_names.split("&&"): # there can be multiple hits for single query sequence
# Get name of result FASTQ file to write this read in
binned_file_path = os.path.join(outdir_path, "{}.fast5".format(hit_name))
if binned_file_path not in srt_file_dict.keys():
srt_file_dict = update_file_dict(srt_file_dict, binned_file_path)
# end if
f5_cpy_func(from_f5, read_name, srt_file_dict[binned_file_path])
# end for
seqs_pass += 1
# end if
# end for
from_f5.close()
# Close all binned files
for file_obj in filter(lambda x: not x is None, srt_file_dict.values()):
file_obj.close()
# end for
sys.stdout.write('\r')
printlog_info_time("File `{}` is binned.".format(os.path.basename(f5_path)))
printn(" Working...")
return (seqs_pass, QL_seqs_fail, align_seqs_fail)