def customize(self, config, reference, input_files, output_file, directory,
dependencies=()):
input_files = safe_coerce_to_tuple(input_files)
stats_out_fname = "Stats_out_MCMC_correct_prob.csv"
command = AtomicCmdBuilder(["mapDamage", "--rescale-only",
"-i", "%(TEMP_IN_BAM)s",
"-d", "%(TEMP_DIR)s",
"-r", "%(IN_REFERENCE)s",
"--rescale-out", "%(OUT_BAM)s"],
TEMP_IN_BAM=MultiBAMInputNode.PIPE_FILE,
IN_REFERENCE=reference,
TEMP_OUT_LOG="Runtime_log.txt",
TEMP_OUT_CSV=stats_out_fname,
OUT_BAM=output_file,
CHECK_VERSION=MAPDAMAGE_VERSION)
command.add_multiple_kwargs(input_files)
return {"command": command,
"config": config,
"input_files": input_files,
"directory": directory,
"dependencies": dependencies}
def customize(self,
config,
reference,
input_files,
output_file,
directory,
dependencies=()):
input_files = safe_coerce_to_tuple(input_files)
stats_out_fname = "Stats_out_MCMC_correct_prob.csv"
command = AtomicCmdBuilder([
"mapDamage", "--rescale-only", "-i", "%(TEMP_IN_BAM)s", "-d",
"%(TEMP_DIR)s", "-r", "%(IN_REFERENCE)s", "--rescale-out",
"%(OUT_BAM)s"
],
TEMP_IN_BAM=MultiBAMInputNode.PIPE_FILE,
IN_REFERENCE=reference,
TEMP_OUT_LOG="Runtime_log.txt",
TEMP_OUT_CSV=stats_out_fname,
OUT_BAM=output_file,
CHECK_VERSION=MAPDAMAGE_VERSION)
command.add_multiple_kwargs(input_files)
return {
"command": command,
"config": config,
"input_files": input_files,
"directory": directory,
"dependencies": dependencies
}
def test_builder__add_multiple_kwargs_multiple_times():
expected = {"IN_FILE_01": "file_a", "IN_FILE_02": "file_b"}
builder = AtomicCmdBuilder("ls")
kwargs = builder.add_multiple_kwargs(("file_a",))
assert_equal(kwargs, {"IN_FILE_01": "file_a"})
kwargs = builder.add_multiple_kwargs(("file_b",))
assert_equal(kwargs, {"IN_FILE_02": "file_b"})
assert_equal(builder.kwargs, expected)
assert_equal(builder.call, ["ls"])
def test_builder__add_multiple_kwargs_multiple_times():
expected = {"IN_FILE_01": "file_a", "IN_FILE_02": "file_b"}
builder = AtomicCmdBuilder("ls")
kwargs = builder.add_multiple_kwargs(("file_a", ))
assert_equal(kwargs, {"IN_FILE_01": "file_a"})
kwargs = builder.add_multiple_kwargs(("file_b", ))
assert_equal(kwargs, {"IN_FILE_02": "file_b"})
assert_equal(builder.kwargs, expected)
assert_equal(builder.call, ["ls"])
def customize(self,
config,
reference,
input_files,
output_directory,
title="mapDamage",
dependencies=()):
input_files = safe_coerce_to_tuple(input_files)
command = AtomicCmdBuilder(
[
"mapDamage",
"--no-stats",
# Prevent references with many contigs from using excessive
# amounts of memory, at the cost of per-contig statistics:
"--merge-reference-sequences",
"-t",
title,
"-i",
"%(TEMP_IN_BAM)s",
"-d",
"%(TEMP_DIR)s",
"-r",
"%(IN_REFERENCE)s"
],
TEMP_IN_BAM=MultiBAMInputNode.PIPE_FILE,
IN_REFERENCE=reference,
OUT_FREQ_3p=os.path.join(output_directory, "3pGtoA_freq.txt"),
OUT_FREQ_5p=os.path.join(output_directory, "5pCtoT_freq.txt"),
OUT_COMP_USER=os.path.join(output_directory, "dnacomp.txt"),
OUT_PLOT_FRAG=os.path.join(output_directory,
"Fragmisincorporation_plot.pdf"),
OUT_PLOT_LEN=os.path.join(output_directory, "Length_plot.pdf"),
OUT_LENGTH=os.path.join(output_directory, "lgdistribution.txt"),
OUT_MISINCORP=os.path.join(output_directory,
"misincorporation.txt"),
OUT_LOG=os.path.join(output_directory, "Runtime_log.txt"),
TEMP_OUT_STDOUT="pipe_mapDamage.stdout",
TEMP_OUT_STDERR="pipe_mapDamage.stderr",
CHECK_RSCRIPT=RSCRIPT_VERSION,
CHECK_MAPDAMAGE=MAPDAMAGE_VERSION)
command.add_multiple_kwargs(input_files)
return {
"command": command,
"config": config,
"input_files": input_files,
"dependencies": dependencies
}
def test_builder__add_multiple_kwargs_with_template():
values = ("file_a", "file_b")
expected = {"OUT_BAM_1": "file_a", "OUT_BAM_2": "file_b"}
builder = AtomicCmdBuilder("ls")
kwargs = builder.add_multiple_kwargs(values, template="OUT_BAM_%i")
assert_equal(kwargs, expected)
assert_equal(builder.kwargs, expected)
assert_equal(builder.call, ["ls"])
def test_builder__add_multiple_kwargs():
values = ("file_a", "file_b")
expected = {"IN_FILE_01": "file_a", "IN_FILE_02": "file_b"}
builder = AtomicCmdBuilder("ls")
kwargs = builder.add_multiple_kwargs(values)
assert_equal(kwargs, expected)
assert_equal(builder.kwargs, expected)
assert_equal(builder.call, ["ls"])
def test_builder__add_multiple_kwargs_with_template():
values = ("file_a", "file_b")
expected = {"OUT_BAM_1": "file_a", "OUT_BAM_2": "file_b"}
builder = AtomicCmdBuilder("ls")
kwargs = builder.add_multiple_kwargs(values, template="OUT_BAM_%i")
assert_equal(kwargs, expected)
assert_equal(builder.kwargs, expected)
assert_equal(builder.call, ["ls"])
def test_builder__add_multiple_kwargs():
values = ("file_a", "file_b")
expected = {"IN_FILE_01": "file_a", "IN_FILE_02": "file_b"}
builder = AtomicCmdBuilder("ls")
kwargs = builder.add_multiple_kwargs(values)
assert_equal(kwargs, expected)
assert_equal(builder.kwargs, expected)
assert_equal(builder.call, ["ls"])
def customize(self, config, reference, input_files, output_directory,
title="mapDamage", dependencies=()):
input_files = safe_coerce_to_tuple(input_files)
command = AtomicCmdBuilder(
["mapDamage", "--no-stats",
# Prevent references with many contigs from using excessive
# amounts of memory, at the cost of per-contig statistics:
"--merge-reference-sequences",
"-t", title,
"-i", "%(TEMP_IN_BAM)s",
"-d", "%(TEMP_DIR)s",
"-r", "%(IN_REFERENCE)s"],
TEMP_IN_BAM=MultiBAMInputNode.PIPE_FILE,
IN_REFERENCE=reference,
OUT_FREQ_3p=os.path.join(output_directory, "3pGtoA_freq.txt"),
OUT_FREQ_5p=os.path.join(output_directory, "5pCtoT_freq.txt"),
OUT_COMP_USER=os.path.join(output_directory, "dnacomp.txt"),
OUT_PLOT_FRAG=os.path.join(output_directory,
"Fragmisincorporation_plot.pdf"),
OUT_PLOT_LEN=os.path.join(output_directory, "Length_plot.pdf"),
OUT_LENGTH=os.path.join(output_directory, "lgdistribution.txt"),
OUT_MISINCORP=os.path.join(output_directory,
"misincorporation.txt"),
OUT_LOG=os.path.join(output_directory, "Runtime_log.txt"),
TEMP_OUT_STDOUT="pipe_mapDamage.stdout",
TEMP_OUT_STDERR="pipe_mapDamage.stderr",
CHECK_RSCRIPT=RSCRIPT_VERSION,
CHECK_MAPDAMAGE=MAPDAMAGE_VERSION)
command.add_multiple_kwargs(input_files)
return {"command": command,
"config": config,
"input_files": input_files,
"dependencies": dependencies}