def main():
"""Entry point."""
opts = docopt(__doc__)
data = tofile.read_file("%s/config.yaml" % opts["--in"])
meta = yaml.full_load(data)
prefix = meta["assembly"]["prefix"]
blobdir = prefix
revision = meta.get("revision", 0)
if revision:
blobdir += ".%d" % int(revision)
indir = opts["--in"]
outdir = "%s/%s" % (opts["--out"], prefix)
bindir = opts.get("--bin", None)
Path(outdir).mkdir(parents=True, exist_ok=True)
create_pipeline_directory(indir, "%s/%s.pipeline" % (indir, prefix))
transfer_files(r"%s/*.tar" % indir, outdir)
transfer_files(r"%s/*.gz" % indir, outdir)
transfer_files("%s/%s.pipeline" % (indir, prefix), outdir, compress=True)
if Path("%s/view" % indir).is_dir():
create_static_directory(
"%s/view/%s" % (indir, blobdir), "%s/%s.static" % (indir, blobdir)
)
transfer_files("%s/view/%s" % (indir, blobdir), outdir, compress=True)
transfer_files("%s/%s.static" % (indir, blobdir), outdir)
else:
transfer_files("%s/blobtools/%s" % (indir, blobdir), outdir, compress=True)
transfer_files(r"%s/blastn/%s.*.out" % (indir, prefix), outdir, compress=True)
transfer_files("%s/blobtools/%s.meta.yaml" % (indir, prefix), outdir, compress=True)
transfer_files(r"%s/busco/%s.busco.*_odb10" % (indir, prefix), outdir)
transfer_files("%s/config.yaml" % indir, outdir)
transfer_files(
"%s/cov_stats/%s.chunk_stats.tsv" % (indir, prefix), outdir, compress=True
)
transfer_files(
r"%s/diamond/%s.*.out" % (indir, prefix),
outdir,
rename=("diamond.", "diamond_blastx."),
compress=True,
)
transfer_files(
"%s/diamond/%s.fasta.chunks" % (indir, prefix),
outdir,
rename=("fasta.chunks", "chunks.fasta"),
compress=True,
)
transfer_files(
r"%s/diamond_blastp/%s.*.out" % (indir, prefix),
outdir,
rename=("diamond.", "diamond_blastp."),
compress=True,
)
transfer_files(r"%s/minimap/%s.*.bam" % (indir, prefix), outdir)
transfer_files("%s/pipeline/%s" % (indir, prefix), outdir, compress=True)
transfer_files(
"%s/window_stats/%s.window_stats.tsv" % (indir, prefix), outdir, compress=True
)
transfer_files(
"%s/windowmasker/%s.windowmasker.fasta" % (indir, prefix), outdir, compress=True
)
remove_unwanted_files(indir, bindir)
def remove_unwanted_files(indir, bindir):
"""Move unwanted files to bin or delete."""
LOGGER.info("Removing unwanted files")
if bindir is not None:
LOGGER.info("Moving %s to %s", indir, bindir)
Path(bindir).mkdir(parents=True, exist_ok=True)
shutil.move(indir, bindir)
else:
LOGGER.info("Deleting %s" % indir)
shutil.rmtree(indir)
if __name__ == "__main__":
opts = docopt(__doc__)
data = tofile.read_file("%s/config.yaml" % opts["--in"])
meta = yaml.full_load(data)
prefix = meta["assembly"]["prefix"]
blobdir = prefix
revision = meta.get("revision", 0)
if revision:
blobdir += ".%d" % int(revision)
indir = opts["--in"]
outdir = "%s/%s" % (opts["--out"], prefix)
bindir = opts.get("--bin", None)
Path(outdir).mkdir(parents=True, exist_ok=True)
create_pipeline_directory(indir, "%s/%s.pipeline" % (indir, prefix))
transfer_files(r"%s/*.tar" % indir, outdir)
transfer_files(r"%s/*.gz" % indir, outdir)
transfer_files("%s/%s.pipeline" % (indir, prefix), outdir, compress=True)
if Path("%s/view" % indir).is_dir():
def test_read_file_that_does_not_exist():
"""Test read_file with non-existent files."""
assert tofile.read_file("nofile") is None
assert tofile.read_file("nofile.gz") is None
assert tofile.read_file("tests/files/nofile") is None
assert tofile.read_file("tests/files/nofile.gz") is None
def test_read_file_that_exists():
"""Test read_file with a valid files."""
assert tofile.read_file("tests/files/infile") == "testfile content\n"
assert tofile.read_file("tests/files/infile.gz") == "testfile content\n"
project_list = set()
for child_bioproject in children:
sleep(0.5)
accession = fetch_accession(child_bioproject)
if accession is not None:
accessions.append(accession)
# project_list.add(child_bioproject)
project_list.add(new_projects[child_bioproject])
return accessions, project_list
if __name__ == "__main__":
opts = docopt(__doc__)
bioproject = opts["<BIOPROJECT>"]
outdir = opts["--out"]
os.makedirs(outdir, exist_ok=True)
projects_file = "%s/bioprojects.processed" % outdir
targets_file = "%s/accessions.todo" % outdir
active_file = "%s/accessions.active" % outdir
processed_file = "%s/accessions.processed" % outdir
projects = tofile.read_file(projects_file)
accessions, new_projects = fetch_bioproject_accessions(bioproject,
projects=projects)
with open(targets_file, "a+") as tfh:
tfh.writelines([accession + "\n" for accession in accessions])
with open(projects_file, "a+") as pfh:
pfh.writelines([bioproject + "\n" for bioproject in new_projects])
# TODO: introduce page numbering for when list gets long