def test_checkdir_with_str(temporary_folder):
target_directory = temporary_folder / "target"
# Make sure the target doesn't already exist
assert not target_directory.exists()
target_directory_string = str(target_directory)
utilities.checkdir(target_directory_string)
assert target_directory.exists()
def run(self, output_folder: Path, *reads) -> programio.FastQCOutput:
utilities.checkdir(output_folder)
command = self.get_command(output_folder, reads)
output = self.get_output(output_folder, reads)
if not output.exists():
systemio.command_runner.run(command, output_folder, srun=False)
return output
def sample_variant_calling(reference: Path,
samples: List[sampleio.SampleReads],
project_folder: Path,
ispop: bool = False):
"""
Performs simple variant calling between the supplied reference and the given samples.
Parameters
----------
reference: Path
The sample to use for the project
samples: List[sampleio.SampleReads
Contsins the source reads for variant calling. Trimming is not performmed at this stage.
project_folder: Path
The folder to use for the overall project.
ispop: bool; default False
Whether to run variant calling as populations or clones.
"""
# First validate the input parameters
cancel = not utilities.verify_file_exists(reference)
cancel = cancel or not sampleio.verify_samples(samples)
if cancel:
message = "Something went wrong when validating the variant calling parameters!"
raise ValueError(message)
# Set up the environment
utilities.checkdir(project_folder)
systemio.command_runner.set_command_log(project_folder /
"commandlog_variant_calling.sh")
systemio.command_runner.write_command_to_commandlog(
['module', 'load', 'breseq'])
breseq_workflow = breseq.Breseq(reference, threads=16, population=ispop)
breseq_workflow.test()
results = list()
for index, sample in enumerate(samples):
logger.info(
f"Running variant calling on sample {index} of {len(samples)}: {sample.name}"
)
sample_folder = utilities.checkdir(project_folder / sample.name)
breseq_folder = sample_folder / "breseq"
result = breseq_workflow.run(breseq_folder, sample.forward,
sample.reverse)
results.append(result)
return results
def main_shelly():
logger.info(" Running the large dataset of samples.")
project_folder = Path.home() / "projects" / "shelly"
project_output_folder = utilities.checkdir(project_folder / "output")
table_filename = project_folder / "samples.tsv"
reference = project_folder / "T4.gbff"
samples = sampleio.get_samples_from_table(table_filename)
# Trim the reads
trimmed_output = trim(samples, project_output_folder)
trimmed_samples = [i.as_sample() for i in trimmed_output]
# Call variants
sample_variant_calling(reference,
trimmed_samples,
project_output_folder,
ispop=True)
def process_reference(
self, reference: Union[str, Path, sampleio.SampleReads,
programio.TrimmomaticOutput]
) -> Path:
# If the reference already points to an assembly, return it without modification.
if isinstance(reference, Path) and reference.is_file():
return reference
# Otherwise, need to convert the input to the path to an assembly.
# Since the assembly does not yet exist, the input must be a reference to a set of reads.
# Coerce this reference to either SampleReads or TrimmomaticOutput, which can then be fed through the
# AssemblyWorkflow.
# test whether `reference` is a folder pointing to a set of reads.
if isinstance(reference, Path) and reference.is_dir():
reference = get_reads_from_folder(reference)
# `reference` should now be one of the accepts `Reads` objects.
assembly_workflow = read_assembly.AssemblyWorkflow(self.project_folder)
reference_assembly = assembly_workflow.assemble_sample(
reference,
utilities.checkdir(self.project_folder / reference.name))
return reference_assembly.contigs
def __init__(self, project_folder: Path):
self.project_folder = utilities.checkdir(project_folder)
def reads_trimmed(project_folder) -> Path: source_folder = DATA_FOLDER / "outputs" / "trimmomatic" destination_folder = utilities.checkdir(project_folder / "reads") utilities.copydir(source_folder, destination_folder) return destination_folder
def reads_raw(project_folder) -> Path: source_folder = DATA_FOLDER / "inputs" / "reads" destination_folder = utilities.checkdir(project_folder / "reads") utilities.copydir(source_folder, destination_folder) return destination_folder
def project_folder(tmp_path) -> Path: f = utilities.checkdir(tmp_path / "project") return f
