def clinker(
files,
identity=0.3,
delimiter=None,
decimals=2,
plot=None,
output=None,
force=False,
hide_link_headers=False,
hide_alignment_headers=False,
):
"""Entry point for running the script."""
LOG.info("Starting clinker")
# Check output file before doing anything else
if output and Path(output).exists() and not force:
LOG.error(
f"File {output} already exists but --force not specified, exiting")
return
# Parse files, generate objects
paths = find_files(files)
LOG.info("Parsing GenBank files: %s", paths)
clusters = parse_files(paths)
# Align all clusters
if len(clusters) == 1:
globaligner = align.align_clusters(clusters[0])
else:
LOG.info("Starting cluster alignments")
globaligner = align.align_clusters(*clusters, cutoff=identity)
LOG.info("Generating results summary...")
summary = globaligner.format(
delimiter=delimiter,
decimals=decimals,
link_headers=not hide_link_headers,
alignment_headers=not hide_alignment_headers,
)
if output:
LOG.info(f"Writing alignments to {output}")
with open(output, "w") as fp:
fp.write(summary)
else:
print(summary)
# Generate the SVG
if plot:
LOG.info("Building clustermap.js visualisation")
if isinstance(plot, str):
LOG.info("Writing to: %s", plot)
plot_clusters(globaligner, output=None if plot is True else plot)
LOG.info("Done!")
return globaligner
def clinker(
files,
session=None,
identity=0.3,
delimiter=None,
decimals=2,
plot=None,
output=None,
force=False,
no_align=False,
hide_link_headers=False,
hide_alignment_headers=False,
use_file_order=False,
json_indent=None,
jobs=None,
):
"""Entry point for running the script."""
LOG.info("Starting clinker")
load_session = session and Path(session).exists()
if load_session:
LOG.info("Loading session from: %s", session)
with open(session) as fp:
globaligner = align.Globaligner.from_json(fp)
if files:
paths = find_files(files)
if not paths:
LOG.error("No files found")
raise SystemExit
LOG.info("Parsing GenBank files: %s", paths)
clusters = parse_files(paths)
LOG.info("Adding clusters to loaded session and aligning")
globaligner.add_clusters(*clusters)
globaligner.align_stored_clusters(cutoff=identity, jobs=jobs)
load_session = False
else:
# Parse files, generate objects
paths = find_files(files)
if not paths:
LOG.error("No files found")
raise SystemExit
LOG.info("Parsing GenBank files: %s", paths)
clusters = parse_files(paths)
# Align all clusters
if no_align:
globaligner = align.Globaligner()
globaligner.add_clusters(*clusters)
elif len(clusters) == 1:
globaligner = align.align_clusters(clusters[0], jobs=1)
else:
LOG.info("Starting cluster alignments")
globaligner = align.align_clusters(*clusters, cutoff=identity, jobs=jobs)
LOG.info("Generating results summary...")
summary = globaligner.format(
delimiter=delimiter,
decimals=decimals,
link_headers=not hide_link_headers,
alignment_headers=not hide_alignment_headers,
)
if output:
if (output and Path(output).exists() and not force):
print(summary)
LOG.warn("File %s already exists but --force was not specified", output)
else:
LOG.info("Writing alignments to: %s", output)
with open(output, "w") as fp:
fp.write(summary)
else:
print(summary)
if session and not load_session:
LOG.info("Saving session to: %s", session)
with open(session, "w") as fp:
globaligner.to_json(fp, indent=json_indent)
# Generate the SVG
if plot:
LOG.info("Building clustermap.js visualisation")
if isinstance(plot, str):
LOG.info("Writing to: %s", plot)
plot_clusters(
globaligner,
output=None if plot is True else plot,
use_file_order=use_file_order,
)
LOG.info("Done!")
return globaligner
def clinker(
files,
session=None,
identity=0.3,
delimiter=None,
decimals=2,
plot=None,
output=None,
force=False,
no_align=False,
hide_link_headers=False,
hide_alignment_headers=False,
use_file_order=False,
json_indent=None,
jobs=None,
ranges=None,
matrix_out=None,
gene_functions=None,
):
"""Entry point for running the script."""
LOG.info("Starting clinker")
load_session = session and Path(session).exists()
# Allow no files, so that user can generate a blank clinker web app
# and load in previously saved figure data
if not files:
LOG.info("No files provided!")
if plot:
LOG.info("Opening empty clinker web app...")
plot_data(
dict(clusters=[], links=[], groups=[]),
output=None if plot is True else plot
)
return
# Parse range strings, if any specified
if ranges:
ranges = parse_ranges(ranges)
# Parse any gene functions for grouping, if specified
if gene_functions:
gene_functions = parse_gene_functions(gene_functions)
if load_session:
LOG.info("Loading session from: %s", session)
with open(session) as fp:
globaligner = align.Globaligner.from_json(fp)
if files:
paths = find_files(files)
if not paths:
LOG.error("No files found")
raise SystemExit
LOG.info("Parsing files:")
clusters = parse_files(paths, ranges=ranges)
LOG.info("Adding clusters to loaded session and aligning")
globaligner.add_clusters(*clusters)
globaligner.align_stored_clusters(cutoff=identity, jobs=jobs)
globaligner.build_gene_groups(functions=gene_functions)
load_session = False
else:
# Parse files, generate objects
paths = find_files(files)
if not paths:
LOG.error("No files found")
raise SystemExit
LOG.info("Parsing files:")
clusters = parse_files(paths, ranges=ranges)
# Align all clusters
if no_align:
globaligner = align.Globaligner()
globaligner.add_clusters(*clusters)
globaligner.build_gene_groups(functions=gene_functions)
elif len(clusters) == 1:
globaligner = align.align_clusters(clusters[0], jobs=1)
else:
LOG.info("Starting cluster alignments")
globaligner = align.align_clusters(*clusters, cutoff=identity, jobs=jobs)
globaligner.build_gene_groups(functions=gene_functions)
if globaligner.alignments:
LOG.info("Generating results summary...")
summary = globaligner.format(
delimiter=delimiter,
decimals=decimals,
link_headers=not hide_link_headers,
alignment_headers=not hide_alignment_headers,
)
if output:
if (output and Path(output).exists() and not force):
print(summary)
LOG.warn("File %s already exists but --force was not specified", output)
else:
LOG.info("Writing alignments to: %s", output)
with open(output, "w") as fp:
fp.write(summary)
else:
print(summary)
if matrix_out:
LOG.info("Writing synteny matrix to: %s", matrix_out)
matrix = globaligner.format_matrix(normalise=True, as_distance=True)
with open(matrix_out, "w") as fp:
fp.write(matrix)
else:
LOG.info("No alignments were generated")
if session and not load_session:
LOG.info("Saving session to: %s", session)
with open(session, "w") as fp:
globaligner.to_json(fp, indent=json_indent)
# Generate the SVG
if plot:
LOG.info("Building clustermap.js visualisation")
if isinstance(plot, str):
LOG.info("Writing to: %s", plot)
plot_clusters(
globaligner,
output=None if plot is True else plot,
use_file_order=use_file_order,
)
LOG.info("Done!")
return globaligner