def __init__(self, data: Path or None, config: Path):
PoreLogger.__init__(self, level=logging.INFO, name='DataStorage')
self.data = data
self.workdir = Path.home() / '.npc' / '.storage_client'
self.netflow = NetflowClient(node_config=config,
workdir=Path.home() / '.npc' /
'.storage_client')
self.netflow.configure_storage_nodes(client_entry='storage')
def __init__(self, pipeline_client: PipelineClient):
PoreLogger.__init__(self, level=logging.INFO, name='PipelinePollster')
self.pipeline_client = pipeline_client
# When instantiated, remove the previous registry if present
self.result_registry = self.pipeline_client.workdir / 'result.registry'
# List of files from last poll
self.last_file_poll = []
# If pipeline screen is queried:
self.last_screen_open = False
def __init__(self, name: str, collection: str, version: int, author: str,
data: Path):
PoreLogger.__init__(self, level=logging.INFO, name="Dataset")
self.name = name
self.collection = collection
self.version = version
self.data = data
self.author = author
self.dataset = Path(f"{self.collection}-{self.name}-v{self.version}")
self.files = None
def __init__(self, host, user, ssh_key, name, logger_name: str = 'PipelineClient'):
PoreLogger.__init__(self, level=logging.INFO, name=logger_name)
self.remote = True if host != 'local' else False
self.name = name
self.host = host
self.user = user
self.ssh_key = ssh_key
self.client = None
self.sftp = None
self.conn = None
self.active_screens = {}
def __init__(self, pathogen_path: Path = None, signal_path: Path = None):
PoreLogger.__init__(self, level=logging.INFO, name="AppServer")
self.pipelines = {}
self.pathogen_path = pathogen_path
self.signal_path = signal_path
if self.pathogen_path:
self.pipelines['pathogen'] = PathogenPipeline()
if self.signal_path:
self.pipelines['signal'] = SignalPipeline()
pass
def __init__(self,
path=os.getcwd(),
species="Staphylococcus aureus",
force=False,
verbose=True):
PoreLogger.__init__(self, name="NCTC")
self.date = time.strftime("%d-%m-%Y-%H-%M")
self.path = os.path.abspath(path)
self.project = True
self.species = species
self.species_name = self.species.replace(" ", "_").lower()
self.species_path = os.path.join(self.path, self.species_name)
self.force = force
self.verbose = verbose
# Infrastructure
self.config_path = os.path.join(self.path, ".nctc")
# Analysis with NCTC Programs and Snakemake
self.nctc = os.path.dirname(os.path.dirname(
os.path.realpath(__file__)))
self.abricate = os.path.join(self.nctc, "bin", "abricate", "bin",
"abricate")
self.env_file = os.path.join(self.nctc, "env", "nctc.json")
self.snake_file = os.path.join(self.nctc, "pipe", "nctc.snk")
self.config_file = os.path.join(self.nctc, "pipe", "nctc.json")
self.config_run_file = None
# HTML NCTC3000
self.url = "http://www.sanger.ac.uk/resources/downloads/bacteria/nctc/"
self.tables = dict()
self.files = dict()
self.report_files = dict()
def __init__(self, app_data: dict, db_reports: dict, group_reads: dict):
PoreLogger.__init__(self, level=logging.INFO, name="PathogenReport")
self.app_data = app_data # report configuration sent by app
self.db_reports = db_reports # database report tables
self.group_reads = group_reads # total reads by group / sample
self.config = dict()
self.report_data = None
self.template_path = Path(__file__).parent / 'templates' / 'reports'
self.sample_id = self.app_data['header']['sample_id']
self.uuid = str(uuid4())
self.uid = self.uuid.split('-')[0]
def __init__(self, node_config: Path, workdir: Path):
PoreLogger.__init__(self, level=logging.INFO, name='NanoPathClient')
logging.getLogger("paramiko").setLevel(logging.WARNING)
self.logger.info(
f"{Y}NanoPathClient{RE}"
)
# Read the netflow configurations
self.config = self.read_config(file=node_config)
# Working directory structure
self.workdir = workdir
self.nexus = self.workdir / 'nexus'
self.nexus.mkdir(parents=True, exist_ok=True)
# Store the clients for later use:
self.clients = {}
def remove_human(fq, fq1, fq2, ref, out, mapq, preset, human_out, threads):
""" Filter a read file (single, paired) by mapping to human reference index """
logger = PoreLogger(level=logging.INFO, name="HostRemoval").logger
remove_by_alignment(fq=fq,
ref=ref,
out=out,
mapq=mapq,
preset=preset,
human_out=human_out,
threads=threads,
logger=logger)
def rename_barcodes(directory, file, panel, outdir, extension, megalodon):
""" Rename barcode read files after basecalling"""
pl = PoreLogger(level=logging.INFO, name="Utils").logger
pl.info(f'Creating output directory: {outdir}')
outdir.mkdir(parents=True, exist_ok=True)
pl.info(f'Reading panel data from file: {file}')
df = pandas.read_csv(file, sep='\t', header=0)
if panel:
pl.info(f'Subsetting panel data to: {panel}')
df = df.loc[df['panel'].isin([str(panel)]), :]
if megalodon:
df['megalodon'] = df[['panel', 'barcode']].apply(lambda x: '_'.join(x),
axis=1)
for d in directory.glob("*"):
if d.name in df.megalodon.tolist():
row = df.loc[df.megalodon == d.name, :]
name = row.name.values[0]
pl.info(f"Copy directory {d.name} to {outdir / name}")
shutil.copytree(str(d), str(outdir / name))
else:
for p in directory.rglob("*"):
if p.stem in df.barcode.tolist():
row = df.loc[df.barcode == p.stem, :]
name = row.name.values[0] + f"{extension}"
pl.info(f'Rename and copy {p.name} to: {outdir / name}')
shutil.copy(str(p), str(outdir / name))