def dict_file(self, options):
"""Generate GTDB dictionary file."""
check_file_exists(options.taxonomy_file)
make_sure_path_exists(options.output_dir)
p = WebsiteData(options.release_number, options.output_dir)
p.dict_file(options.taxonomy_file)
self.logger.info('Done.')
def gene_files(self, options):
"""Generate untrimmed gene files for bac120/ar122"""
make_sure_path_exists(options.output_dir)
check_file_exists(options.user_gid_table)
w = WebsiteData(options.release_number, options.output_dir)
w.gene_files(options.user_gid_table, options.taxonomy_file,
options.cpus)
self.logger.info('Done.')
def nucleotide_files(self, options):
"""Generate the archive containing all protein (AA) for representative genomes."""
make_sure_path_exists(options.output_dir)
check_file_exists(options.user_gid_table)
check_file_exists(options.genome_dirs)
w = WebsiteData(options.release_number, options.output_dir)
w.nucleotide_files(options.user_gid_table, options.taxonomy_file,
options.genome_dirs, options.uba_path)
self.logger.info('Done.')
def hq_genome_file(self, options):
"""Generate file indicating HQ genomes."""
check_file_exists(options.metadata_file)
if options.user_gid_table.lower() != 'none':
check_file_exists(options.user_gid_table)
make_sure_path_exists(options.output_dir)
p = WebsiteData(options.release_number, options.output_dir)
p.hq_genome_file(options.metadata_file, options.user_gid_table)
self.logger.info('Done.')
def sp_cluster_file(self, options):
"""Generate file indicating GTDB species clusters."""
check_file_exists(options.metadata_file)
check_file_exists(options.gtdb_sp_clusters_file)
check_file_exists(options.user_gid_table)
make_sure_path_exists(options.output_dir)
p = WebsiteData(options.release_number, options.output_dir)
p.sp_cluster_file(options.metadata_file, options.gtdb_sp_clusters_file,
options.user_gid_table)
self.logger.info('Done.')
def taxonomy_files(self, options):
"""Generate taxonomy files for GTDB website."""
check_file_exists(options.metadata_file)
check_file_exists(options.gtdb_sp_clusters_file)
check_file_exists(options.user_gid_table)
make_sure_path_exists(options.output_dir)
p = WebsiteData(options.release_number, options.output_dir)
p.taxonomy_files(options.metadata_file, options.gtdb_sp_clusters_file,
options.user_gid_table)
self.logger.info('Done.')
def tax_comp_files(self, options):
"""Generate comparison files between 2 taxonomy ordered by degree of polyphyly."""
check_file_exists(options.reference_taxonomy_file)
check_file_exists(options.new_taxonomy_file)
make_sure_path_exists(options.output_dir)
p = WebsiteData('', options.output_dir)
p.tax_comp_files(options.reference_taxonomy_file,
options.new_taxonomy_file, options.user_gid_table,
options.changes_only, options.filter_placeholder_name,
options.top_change)
self.logger.info('Done.')
def marker_files(self, options):
"""Generate marker gene file."""
check_dir_exists(options.bac120_gene_dir)
check_dir_exists(options.ar122_gene_dir)
check_file_exists(options.user_gid_table)
make_sure_path_exists(options.output_dir)
p = WebsiteData(options.release_number, options.output_dir)
p.marker_files(options.bac120_gene_dir, options.ar122_gene_dir,
options.user_gid_table)
self.logger.info('Done.')
def tree_files(self, options):
"""Generate tree files spanning representative genomes."""
check_file_exists(options.metadata_file)
check_file_exists(options.bac_tree)
check_file_exists(options.ar_tree)
check_file_exists(options.user_gid_table)
make_sure_path_exists(options.output_dir)
p = WebsiteData(options.release_number, options.output_dir)
p.tree_files(options.metadata_file, options.bac_tree, options.ar_tree,
options.user_gid_table)
self.logger.info('Done.')
def msa_files(self, options):
"""Generate concatenated MSA files."""
check_file_exists(options.bac120_msa_file)
check_file_exists(options.ar122_msa_file)
check_file_exists(options.metadata_file)
check_file_exists(options.user_gid_table)
make_sure_path_exists(options.output_dir)
p = WebsiteData(options.release_number, options.output_dir)
p.msa_files(options.bac120_msa_file, options.ar122_msa_file,
options.metadata_file, options.user_gid_table)
self.logger.info('Done.')
def ssu_files(self, options):
"""Generate 16S rRNA FASTA files for GTDB website."""
check_file_exists(options.metadata_file)
check_file_exists(options.gtdb_sp_clusters_file)
check_file_exists(options.user_gid_table)
check_file_exists(options.genome_path_file)
make_sure_path_exists(options.output_dir)
p = WebsiteData(options.release_number, options.output_dir)
p.ssu_files(options.metadata_file, options.gtdb_sp_clusters_file,
options.user_gid_table, options.genome_path_file,
options.min_16S_ar_len, options.min_16S_bac_len,
options.min_contig_len)
self.logger.info('Done.')
def validate(self, options):
"""Perform validation checks on GTDB website files."""
check_file_exists(options.taxonomy_file)
check_file_exists(options.tree_file)
check_file_exists(options.metadata_file)
check_file_exists(options.msa_file)
check_file_exists(options.ssu_file)
check_file_exists(options.sp_clusters_file)
check_file_exists(options.hq_genome_file)
p = WebsiteData(None, None)
p.validate(options.taxonomy_file, options.tree_file,
options.metadata_file, options.msa_file, options.ssu_file,
options.sp_clusters_file, options.hq_genome_file)
self.logger.info('Done.')
def arb_files(self, options):
"""Generate ARB metadata file."""
self.logger.warning('IN PROGRESS')
return
check_file_exists(options.metadata_file)
check_file_exists(options.metadata_fields)
check_file_exists(options.user_gid_table)
check_file_exists(options.bac120_msa_file)
check_file_exists(options.ar122_msa_file)
make_sure_path_exists(options.output_dir)
p = WebsiteData(options.release_number, options.output_dir)
p.arb_files(options.metadata_file, options.metadata_fields,
options.user_gid_table, options.bac120_msa_file,
options.ar122_msa_file)
self.logger.info('Done.')
def json_tree_parser(self, options):
"""generate Json file used to create the tree in http://gtdb.ecogenomic.org/tree"""
p = WebsiteData(options.release_number, options.output_dir)
p.json_tree_parser(options.taxonomy_file, options.metadata_file)
self.logger.info('Done.')