def write_marker_file(self, sample_barcode):
"""Writes a marker file for the given results"""
marker_folder_id = drive.get_folder_id_of_path(
self.drive_service, self.cfg.ACCESSSION_TRACKING_MARKERS_FOLDER)
with drive.put_file(self.drive_service, marker_folder_id,
sample_barcode):
...
def create_layout_pdf(cfg: Config, entry_data: Dict[str, str]):
"""Main function to read a layout file and write the resulting plate layout map.
Parameters
----------
cfg: Config
configuration information
entry_data: Dict[str, str]
dictionary containing the response that was submitted to Sample Plate Metada.
The required keys are: the researcher, timestamp, sample plate barcode, and a
link to the sample plate map in Google Drive. Optionally, the "local_run" key
is used as a flag to indicate the script is being run from the command line
rather than on AWS.
"""
sample_barcode = entry_data[SampleMetadata.SAMPLE_PLATE_BARCODE]
output_filename = f"{sample_barcode}.pdf"
if LOCAL_RUN in entry_data:
output_path, drive_service = entry_data[LOCAL_RUN]
output_file_object = (output_path / output_filename).open("wb")
else:
logger.debug("getting gdrive credentials")
google_creds = gutils.get_secrets_manager_credentials()
drive_service = drive.get_service(google_creds)
processed_layout_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.LAYOUT_PDF_FOLDER)
output_file_object = drive.put_file(
drive_service,
processed_layout_folder_id,
output_filename,
binary=True,
)
try:
plate_map_file = drive.get_layout_file_from_url(
drive_service, entry_data[SampleMetadata.SAMPLE_PLATE_MAP])
except KeyError:
raise BadDriveURL(
f"Bad URL in {SampleMetadata.SHEET_NAME} for {sample_barcode}")
plate_map_type = accession.get_plate_map_type_from_name(
plate_map_file.name)
with plate_map_file.open() as fh:
accession_data = accession.read_accession_data(plate_map_type, fh)
logger.info(f"Writing layout map to {output_filename}")
with output_file_object as output_fh:
format_pdf(
entry_data[SampleMetadata.SAMPLE_PLATE_BARCODE],
accession_data,
entry_data[SampleMetadata.RESEARCHER_NAME],
format_time(cfg, entry_data[SampleMetadata.TIMESTAMP]),
output_fh,
)
def test_mkdir(
gdrive_service: DriveService,
gdrive_folder: DriveObject,
new_folder_name="mkdir-test",
):
subdir = mkdir(gdrive_service, gdrive_folder.id, new_folder_name)
traversed_folder_id = get_folder_id_of_path(
gdrive_service, [gdrive_folder.name, new_folder_name]
)
assert subdir.id == traversed_folder_id
def __init__(self, drive_service, cfg):
self.completed_pcr_barcodes = get_completed_pcr_barcodes(
drive_service, cfg)
self.results_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.CSV_RESULTS_FOLDER_TRACKING)
self.accession_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.PLATE_LAYOUT_FOLDER)
self.accession_locations = get_accession_locations(drive_service, cfg)
self.mark_as_processed_sample_barcodes = get_mark_as_processed_sample_barcodes(
drive_service, cfg)
self.accessions_sheet = cfg["DATA"].get("accession_tracking_sheet")
self.clin_lab_sheet = cfg["DATA"].get("clin_lab_reporting_sheet")
self.supervisor_plate_queue_sheet = cfg["DATA"].get(
"supervisor_plate_queue_sheet")
form_responses = CollectiveForm(
drive_service, cfg["DATA"].get("collection_form_spreadsheet_id"))
self.registered_df = form_responses[SampleRegistration.SHEET_NAME]
self.bravo_rna_df = form_responses[BravoRNAExtraction.SHEET_NAME]
self.check_in_df = form_responses[FridgeCheckin.SHEET_NAME]
self.starting_bravo_df = form_responses[BravoStart.SHEET_NAME]
self.freezer_check_in_df = form_responses[FreezerCheckin.SHEET_NAME]
def get_already_tracked_samples(drive_service, cfg):
"""Return a list of processed qPCR barcodes by checking our marker file folder"""
marker_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.ACCESSSION_TRACKING_MARKERS_FOLDER)
marker_folder_contents = drive.get_contents_by_folder_id(drive_service,
marker_folder_id,
only_files=True)
completed_barcodes = set(marker_folder_entry.name
for marker_folder_entry in marker_folder_contents)
return completed_barcodes
def __init__(
self,
session: Session,
drive_service: drive.DriveService,
folder_path_components: List[str] = None,
):
super().__init__(session=session)
self.session = session
self.drive_service = drive_service
self.folder_path_components = folder_path_components
self.folder_id = drive.get_folder_id_of_path(
self.drive_service, self.folder_path_components)
self.data = self.initialize_data_from_source()
def test_mkdir_recursive(
gdrive_service: DriveService,
gdrive_folder: DriveObject,
new_folder_name="mkdir-recursive",
):
new_path_components = [new_folder_name, "abc", "abc", "def"]
full_path = [gdrive_folder.name] + new_path_components
subdir_id = mkdir_recursive(gdrive_service, gdrive_folder.id, new_path_components)
lookup = get_folder_id_of_path(gdrive_service, full_path)
assert subdir_id == lookup
second_subdir_id = mkdir_recursive(
gdrive_service, gdrive_folder.id, new_path_components
)
assert second_subdir_id == lookup
def fetch_barcodes(args, cfg):
google_credentials = gutils.get_secrets_manager_credentials(args.secret_id)
drive_service = drive.get_service(google_credentials)
# qpcr logs folder
logs_folder_id = drive.get_folder_id_of_path(drive_service,
cfg.PCR_LOGS_FOLDER)
logs_folder_contents = drive.get_contents_by_folder_id(drive_service,
logs_folder_id,
only_files=True)
barcodes_to_fetch = defaultdict(RunFiles)
for entry in logs_folder_contents:
m = RunFiles.get_qpcr_file_type(entry.name)
if m is None:
continue
elif m[RunFiles.BARCODE] in args.barcodes:
barcodes_to_fetch[m[RunFiles.BARCODE]].add_file(m, entry)
for barcode, barcode_files in barcodes_to_fetch.items():
# all files must be present, at least one quant_amp file
if not barcode_files.all_files:
logger.warning(msg=f"Missing files for {barcode}!")
continue
logger.info(msg=f"Found sample to fetch: {barcode}")
# read in the run information and quant cq
run_info = barcode_files.run_info
logger.info(msg=f" Downloading: {run_info.name}")
with drive.get_file(drive_service, run_info.id, binary=False) as fh:
with (args.output_dir / run_info.name).open("w") as out:
out.write(fh.read())
quant_cq = barcode_files.quant_cq
logger.info(msg=f" Downloading: {quant_cq.name}")
with drive.get_file(drive_service, quant_cq.id, binary=False) as fh:
with (args.output_dir / quant_cq.name).open("w") as out:
out.write(fh.read())
for quant_amp in barcode_files.quant_amp.values():
logger.info(msg=f" Downloading: {quant_amp.name}")
with drive.get_file(drive_service, quant_amp.id,
binary=False) as fh:
with (args.output_dir / quant_amp.name).open("w") as out:
out.write(fh.read())
def get_accession_locations(drive_service, cfg) -> Dict[str, Tuple[str, str]]:
"""return a mapping between accession ID's and their origin location"""
accession_locations = {}
accession_location_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.ACCESSION_LOCATIONS_FOLDER)
accession_location_files = drive.get_contents_by_folder_id(
drive_service, accession_location_folder_id, only_files=True)
for accession_location_file in accession_location_files:
with drive.get_file(drive_service, accession_location_file.id) as fh:
accession_location_reader = csv.reader(fh, delimiter=",")
for row in accession_location_reader:
if row[0] == "Accession":
# header row
continue
submitter_id = ""
if len(row) == 3:
accession, location, submitter_id = row
else:
accession, location = row
accession_locations[accession] = location, submitter_id
return accession_locations
def test_get_folder_id_of_path_one_level(gdrive_service: DriveService,
gdrive_folder: DriveObject):
"""Tests that get_folder_id_of_path works with an unnested folder."""
retrieved_folder_id = get_folder_id_of_path(gdrive_service,
[gdrive_folder.name])
assert retrieved_folder_id == gdrive_folder.id
def processing(cfg: Config, google_credentials: service_account.Credentials):
git_info = get_git_info()
drive_service = drive.get_service(google_credentials)
logger.info(msg=f"Starting processing loop with code version: {git_info}")
# qpcr logs folder
logs_folder_id = drive.get_folder_id_of_path(drive_service,
cfg.PCR_LOGS_FOLDER)
# markers folder
markers_folder_id = drive.get_folder_id_of_path(drive_service,
cfg.PCR_MARKERS_FOLDER)
# csv results folder
csv_results_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.CSV_RESULTS_FOLDER)
# CB rad results folder
cb_report_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.CHINA_BASIN_CSV_REPORTS_FOLDER)
# final reports folder
final_results_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.FINAL_REPORTS_FOLDER)
# get the collection spreadsheet
collective_form = CollectiveForm(
drive_service, cfg["DATA"]["collection_form_spreadsheet_id"])
logs_folder_contents = drive.get_contents_by_folder_id(drive_service,
logs_folder_id,
only_files=True)
marker_folder_contents = drive.get_contents_by_folder_id(drive_service,
markers_folder_id,
only_files=True)
plate_layout_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.PLATE_LAYOUT_FOLDER)
completed_barcodes = set(marker_folder_entry.name
for marker_folder_entry in marker_folder_contents)
sample_metadata_form = collective_form[SampleMetadata.SHEET_NAME]
rerun_form = collective_form[SampleRerun.SHEET_NAME]
# group log file entries by barcode
logger.info(msg="Checking for samples to process")
barcodes_to_process = defaultdict(RunFiles)
for entry in logs_folder_contents:
m = RunFiles.get_qpcr_file_type(entry.name)
if m is None or m[RunFiles.BARCODE] in completed_barcodes:
continue
else:
barcodes_to_process[m[RunFiles.BARCODE]].add_file(m, entry)
for barcode, barcode_files in barcodes_to_process.items():
# all files must be present, at least one quant_amp file
if not barcode_files.all_files:
message = f"Missing files for: {barcode}. Skipping for now"
logger.critical(msg=message, extra={"notify_slack": True})
continue
try:
logger.info(msg=f"Found sample to process, barcode: {barcode}")
logger.info(msg=f"Getting metadata and data for: {barcode}")
bravo_metadata = BravoMetadata.load_from_spreadsheet(
barcode,
collective_form,
)
if bravo_metadata.sop_protocol is None:
message = f"Skipping sample plate: {barcode}, no protocol"
logger.critical(msg=message, extra={"notify_slack": True})
continue
protocol = get_protocol(bravo_metadata.sop_protocol)
if not set(barcode_files.quant_amp).issuperset(protocol.mapping):
missing = map(
str,
set(protocol.mapping) - set(barcode_files.quant_amp))
message = f"Missing quant amp files for {barcode}: {', '.join(missing)}"
logger.critical(msg=message, extra={"notify_slack": True})
continue
# process well data and check controls, return results
logger.info(
msg=f"Processing well data and controls for: {barcode}")
accession_data = accession.get_accession_data_with_rerun(
drive_service,
plate_layout_folder_id,
sample_metadata_form,
rerun_form,
bravo_metadata.sample_barcode,
)
control_wells = get_control_wells_from_type(
controls_type=bravo_metadata.controls_type,
accession_data=accession_data,
)
update_accession_data_with_controls(control_wells, accession_data,
barcode)
processing_results = process_barcode(
cfg,
barcode,
barcode_files,
bravo_metadata,
protocol,
control_wells,
accession_data,
)
with drive.put_file(
drive_service,
csv_results_folder_id,
processing_results.results_filename,
) as fh:
processing_results.write_results(fh)
china_basin_result_file = drive.put_file(
drive_service,
cb_report_folder_id,
processing_results.cb_report_filename,
)
with china_basin_result_file as fh:
processing_results.write_cb_report(fh)
# create pdf report
logger.info(
msg=f"Generating and uploading results PDF for: {barcode}")
final_pdf = io.BytesIO()
create_final_pdf(processing_results, final_pdf)
pdf_results_file = drive.put_file(
drive_service,
final_results_folder_id,
processing_results.final_pdf_filename,
)
with pdf_results_file as out_fh:
out_fh.write(final_pdf.getvalue())
logger.info(msg=f"Sending email report: {barcode}")
mail.send_email(
google_credentials,
sender=cfg["EMAIL"].get("sender"),
recipients=cfg["EMAIL"].get("recipients"),
subject=_format_email_subject(
sample_barcode=bravo_metadata.sample_barcode,
qpcr_barcode=barcode,
),
body=_format_email_body(
sample_barcode=bravo_metadata.sample_barcode,
results_file_id=china_basin_result_file.id,
),
attachments={processing_results.final_pdf_filename: final_pdf},
)
message = (
f"Processed sample plate: {bravo_metadata.sample_barcode}-{barcode}"
f" using rev {git_info}")
logger.critical(msg=message, extra={"notify_slack": True})
# write a marker so we don't process this file again.
processing_results.write_marker_file(drive_service,
markers_folder_id)
except Exception as err:
logger.critical(f"Error in [{cfg.aws_env}]: {err}",
extra={"notify_slack": True})
logger.exception("Details:")
def parse_qpcr_csv(args):
cfg = Config()
create_logger(cfg, debug=args.debug)
logger.info(msg=f"Started local processing in: {args.qpcr_run_path}")
if args.use_gdrive and not args.barcodes:
raise ValueError(
"You must specify barcodes to process from Google Drive")
run_path = pathlib.Path(args.qpcr_run_path)
google_credentials = gutils.get_secrets_manager_credentials(args.secret_id)
drive_service = drive.get_service(google_credentials)
collective_form = CollectiveForm(
drive_service, cfg["DATA"]["collection_form_spreadsheet_id"])
sample_metadata_form = collective_form[SampleMetadata.SHEET_NAME]
rerun_form = collective_form[SampleRerun.SHEET_NAME]
if args.use_gdrive:
logs_folder_id = drive.get_folder_id_of_path(drive_service,
cfg.PCR_LOGS_FOLDER)
logs_folder_contents = [
drive_file for drive_file in drive.get_contents_by_folder_id(
drive_service, logs_folder_id, only_files=True)
]
plate_layout_folder_id = drive.get_folder_id_of_path(
drive_service, cfg.PLATE_LAYOUT_FOLDER)
else:
logs_folder_contents = run_path.glob("*.csv")
barcodes_to_process = defaultdict(RunFiles)
for run_file in logs_folder_contents:
m = RunFiles.get_qpcr_file_type(run_file.name)
if m is None:
continue
elif args.barcodes and m[RunFiles.BARCODE] not in args.barcodes:
continue
else:
barcodes_to_process[m[RunFiles.BARCODE]].add_file(m, run_file)
for barcode, barcode_files in barcodes_to_process.items():
# all files must be present, at least one quant_amp file
if not barcode_files.all_files:
message = f"Missing files for: {barcode}. Skipping for now"
logger.info(msg=message)
continue
logger.info(msg=f"Found sample to process, barcode: {barcode}")
logger.info(msg=f"Getting metadata and data for: {barcode}")
bravo_metadata = BravoMetadata.load_from_spreadsheet(
barcode, collective_form)
if args.protocol is not None:
# user specified the protocol
protocol = get_protocol(args.protocol)
else:
protocol = get_protocol(bravo_metadata.sop_protocol)
if not set(barcode_files.quant_amp).issuperset(protocol.mapping):
missing = map(str,
set(protocol.mapping) - set(barcode_files.quant_amp))
message = f"Missing quant amp files for {barcode}: {', '.join(missing)}"
logger.critical(msg=message)
continue
if args.plate_map_file is not None:
plate_map_type = accession.get_plate_map_type_from_name(
args.plate_map_file.name)
accession_data = accession.read_accession_data(
plate_map_type, args.plate_map_file)
elif args.use_gdrive:
accession_data = accession.get_accession_data_with_rerun(
drive_service,
plate_layout_folder_id,
sample_metadata_form,
rerun_form,
bravo_metadata.sample_barcode,
)
else:
raise ValueError(
"You must provide a plate map file or use Google Drive")
control_wells = get_control_wells_from_type(
controls_type=bravo_metadata.controls_type,
accession_data=accession_data,
)
# check for valid accessions
update_accession_data_with_controls(control_wells, accession_data,
barcode)
# process well data and check controls, return results
logger.info(msg=f"Processing well data and controls for: {barcode}")
processing_results = process_barcode(
cfg,
barcode,
barcode_files,
bravo_metadata,
protocol,
control_wells,
accession_data,
)
with (run_path / processing_results.results_filename).open("w") as fh:
processing_results.write_results(fh)
with (run_path /
processing_results.cb_report_filename).open("w") as fh:
processing_results.write_cb_report(fh)
# create pdf report
logger.info(msg=f"Generating results PDF for: {barcode}")
final_pdf_filename = run_path / processing_results.final_pdf_filename
with open(final_pdf_filename, "wb") as output_file:
create_final_pdf(processing_results, output_file)