def apply_filters(alchemist, table, filters, values=None, verbose=False):
"""Applies MySQL WHERE clause filters using a Filter.
:param alchemist: A connected and fully built AlchemyHandler object.
:type alchemist: AlchemyHandler
:param table: MySQL table name.
:type table: str
:param filters: A list of lists with filter values, grouped by ORs.
:type filters: list[list[str]]
:param groups: A list of supported MySQL column names.
:type groups: list[str]
:returns: filter-Loaded Filter object.
:rtype: Filter
"""
db_filter = Filter(alchemist=alchemist, key=table)
db_filter.key = table
db_filter.values = values
try:
db_filter.add(filters)
except:
print("Please check your syntax for the conditional string: "
f"{filters}")
exit(1)
return db_filter
def get_cds_seqrecords(alchemist, values, data_cache=None, nucleotide=False,
verbose=False):
if data_cache is None:
data_cache = {}
cds_list = parse_feature_data(alchemist, values=values)
db_filter = Filter(alchemist)
db_filter.key = 'gene.GeneID'
if verbose:
print("...Converting SQL data...")
seqrecords = []
for cds in cds_list:
parent_genome = data_cache.get(cds.genome_id)
if parent_genome is None:
parent_genome = get_single_genome(alchemist, cds.genome_id,
data_cache=data_cache)
cds.genome_length = parent_genome.length
cds.set_seqfeature()
db_filter.values = [cds.id]
gene_domains = db_filter.select(CDD_DATA_COLUMNS)
record = flat_files.cds_to_seqrecord(cds, parent_genome,
gene_domains=gene_domains)
seqrecords.append(record)
return seqrecords
def get_acc_id_dict(alchemist):
"""Test helper function to retrieve accessions of database entries.
"""
db_filter = Filter(alchemist=alchemist)
db_filter.key = "phage.PhageID"
db_filter.values = db_filter.build_values()
groups = db_filter.group("phage.Accession")
return groups
def execute_resubmit(alchemist,
revisions_data_dicts,
folder_path,
folder_name,
filters="",
groups=[],
verbose=False):
"""Executes the entirety of the genbank resubmit pipeline.
:param alchemist: A connected and fully built AlchemyHandler object.
:type alchemist: AlchemyHandler
:param revisions_data_dicts: Data dictionaries containing pham/notes data.
:type revisions_data_dicts: list[dict]
:param folder_path: Path to a valid dir for new dir creation.
:type folder_path: Path
:param folder_name: A name for the export folder.
:type folder_name: str
:param verbose: A boolean value to toggle progress print statements.
:type verbose: bool
"""
db_filter = Filter(alchemist=alchemist)
db_filter.key = "gene.PhamID"
db_filter.add(BASE_CONDITIONALS)
if filters != "":
try:
db_filter.add(filters)
except:
print("Please check your syntax for the conditional string:\n"
f"{filters}")
resubmit_columns = db_filter.get_columns(RESUBMIT_COLUMNS)
phams = []
for data_dict in revisions_data_dicts:
phams.append(data_dict["Pham"])
db_filter.values = phams
if verbose:
print("Creating export folder...")
export_path = folder_path.joinpath(folder_name)
export_path = basic.make_new_dir(folder_path, export_path, attempt=50)
conditionals_map = {}
export_db.build_groups_map(db_filter,
export_path,
conditionals_map,
groups=groups,
verbose=verbose)
if verbose:
print("Prepared query and path structure, beginning review export...")
for mapped_path in conditionals_map.keys():
if verbose:
print("Retreiving phage data for pham revisions...")
export_dicts = []
for data_dict in revisions_data_dicts:
if verbose:
print(f"...Retrieving data for pham {data_dict['Pham']}...")
conditionals = conditionals_map[mapped_path]
final_call = data_dict["Final Call"]
if final_call == "Hypothetical Protein":
final_call = ""
conditionals.append(
querying.build_where_clause(alchemist.graph,
f"gene.Notes!={final_call}"))
query = querying.build_select(alchemist.graph,
resubmit_columns,
where=conditionals)
results = querying.execute(alchemist.engine,
query,
in_column=db_filter.key,
values=[data_dict["Pham"]])
for result in results:
format_resubmit_data(result, data_dict["Final Call"])
export_dicts.append(result)
if not export_dicts:
if verbose:
print("'{mapped_path.name}' data selected for resubmision "
"matches selected call; no resubmision exported...")
mapped_path.rmdir()
continue
export_dicts = sorted(export_dicts,
key=lambda export_dict: export_dict["Phage"])
if verbose:
print(f"Writing {CSV_NAME} in {mapped_path.name}...")
file_path = mapped_path.joinpath(CSV_NAME)
basic.export_data_dict(export_dicts,
file_path,
RESUBMIT_HEADER,
include_headers=True)
def execute_review(alchemist, folder_path, folder_name,
review=True, values=[],
filters="", groups=[], sort=[],
g_reports=False, s_report=False,
verbose=False):
"""Executes the entirety of the pham review pipeline.
:param alchemist: A connected and fully built AlchemyHandler object.
:type alchemist: AlchemyHandler
:param folder_path: Path to a valid dir for new dir creation.
:type folder_path: Path
:param folder_name: A name for the export folder.
:type folder_name: str
:param csv_title: Title for an appended csv file prefix.
:type csv_title: str
:param review: A boolean to toggle filtering of phams by pham discrepancies.
:type review: bool
:param values: List of values to filter database results.
:type values: list[str]
:param filters: A list of lists with filter values, grouped by ORs.
:type filters: list[list[str]]
:param groups: A list of supported MySQL column names to group by.
:type groups: list[str]
:param sort: A list of supported MySQL column names to sort by.
:param g_reports: A boolean to toggle export of additional pham information.
:type g_reports: bool
:param verbose: A boolean value to toggle progress print statements.
:type verbose: bool
"""
db_filter = Filter(alchemist=alchemist)
db_filter.key = ("gene.PhamID")
if values:
db_filter.values = values
if verbose:
print(f"Identified {len(values)} phams to review...")
if filters != "":
try:
db_filter.add(filters)
except:
print("Please check your syntax for the conditional string:\n"
f"{filters}")
sys.exit(1)
finally:
db_filter.update()
db_filter._filters = []
db_filter._updated = False
db_filter._or_index = -1
db_filter.add(BASE_CONDITIONALS)
db_filter.update()
if not db_filter.values:
print("Current settings produced no database hits.")
sys.exit(1)
if review:
review_phams(db_filter, verbose=verbose)
if sort:
db_filter.sort(sort)
if verbose:
print("Creating export folder...")
export_path = folder_path.joinpath(folder_name)
export_path = basic.make_new_dir(folder_path, export_path, attempt=50)
conditionals_map = {}
export_db.build_groups_map(db_filter, export_path, conditionals_map,
groups=groups,
verbose=verbose)
if verbose:
print("Prepared query and path structure, beginning review export...")
original_phams = db_filter.values
total_g_data = {}
for mapped_path in conditionals_map.keys():
conditionals = conditionals_map[mapped_path]
db_filter.values = original_phams
db_filter.values = db_filter.build_values(where=conditionals)
pf_data = get_pf_data(alchemist, db_filter, verbose=verbose)
write_report(pf_data, mapped_path, PF_HEADER,
csv_name=f"FunctionReport",
verbose=verbose)
if g_reports:
execute_g_report_export(alchemist, db_filter, mapped_path,
total_g_data=total_g_data,
verbose=verbose)
if s_report:
execute_s_report_export(alchemist, db_filter, conditionals,
mapped_path,
verbose=verbose)
