printlog_info('-' * 30)
import src.legacy_taxonomy_handling as legacy_taxonomy_handling
# Configure path to file, which will contain info about what GenBank records to download
acc_fpath = os.path.join(outdir_path, "hits_to_download.tsv")
# Configure path to taxonomy directory
taxonomy_dir = os.path.join(outdir_path, "taxonomy")
if not os.path.isdir(taxonomy_dir):
os.makedirs(taxonomy_dir)
# end if
taxonomy_path = os.path.join(taxonomy_dir, "taxonomy.tsv")
# Check if there is legacy taxonomy file and, if so, reformat it to new (TSV) format
legacy_taxonomy_handling.check_deprecated_taxonomy(outdir_path)
# Dictionary of accessions and record names.
# Accessions are keys, values are tuples: (record_name, number_of_hits).
acc_dict = dict()
# Counter of sequences processed during current run
seqs_processed = [0] # it should be mutable
# Counter of sequences processed concerning possible previous run(s)
glob_seqs_processed = 0
# Varible for stopping execution when probing batch is processed completely.
stop = False
# Further:
for path in fast5_list:
log_info(" {}. `{}`.".format(i, path))
i += 1
# end for
import src.legacy_taxonomy_handling as legacy_taxonomy_handling
# Form path to taxonomy file:
taxonomy_dir = os.path.join(tax_annot_res_dir, "taxonomy")
if not os.path.isdir(taxonomy_dir):
os.makedirs(taxonomy_dir)
# end if
taxonomy_path = os.path.join(taxonomy_dir, "taxonomy.tsv")
# Check if there is legacy taxonomy file and, if so, reformat it to new (TSV) format
legacy_taxonomy_handling.check_deprecated_taxonomy(tax_annot_res_dir)
if n_thr != 1:
from src.spread_files_equally import spread_files_equally
# end if
# |=== Proceed "FAST5-untwisting" if it is enabled ===|
printlog_info('-' * 30)
print()
if untwist_fast5 and not use_old_index:
printlog_info_time("Untwisting started.")
printn(" Working...")