def test_cli_config_rm_motif(tmp_dir):
config_file = os.path.join(tmp_dir, "test_cli_config.motifscanrc")
config = Config(config_file)
config.set_motif_path("motif_set", "/path/to/motif")
config.write()
args = parser.parse_args(["config", "--rm-motif", "motif_set"])
run(args=args, config_file=config_file)
config = Config(config_file)
assert not config.has_motif_set("motif_set")
with pytest.raises(MotifSetNotFoundError):
config.get_motif_path("motif_set")
with pytest.raises(SystemExit):
run(args=args, config_file=config_file)
def test_cli_config_rm_genome(tmp_dir):
config_file = os.path.join(tmp_dir, "test_cli_config.motifscanrc")
config = Config(config_file)
config.set_genome_path("hg19", "/path/to/genome")
config.write()
args = parser.parse_args(["config", "--rm-genome", "hg19"])
run(args=args, config_file=config_file)
config = Config(config_file)
assert not config.has_genome_assembly("hg19")
with pytest.raises(GenomeNotFoundError):
config.get_genome_path("hg19")
with pytest.raises(SystemExit):
run(args=args, config_file=config_file)
def run(args, config_file=None):
if args.list:
config = Config(config_file)
for name, _ in config.list_motif_sets():
print(name)
return
if args.list_remote:
database = JasparDatabase()
if args.database == 'jaspar_core':
remote_sets = database.pfms_core
for name in remote_sets:
print(f"{name:25}\t{database.name + '_CORE'}")
else:
remote_sets = database.pfms_other_collections
for name in remote_sets:
print(f"{name:20}\t{database.name + '_Collections'}")
return
if args.install:
install_motif(args, config_file)
return
if args.build:
build_motif(args, config_file)
return
if args.uninstall:
uninstall_motif(args, config_file)
return
def run(args, config_file=None):
if args.list:
config = Config(config_file)
for name, _ in config.list_genome_assemblies():
print(name)
return
if args.list_remote:
database = UcscDatabase()
for assembly in database.assemblies:
print(f"{assembly.id:12}\t{database.name}\t{assembly.description}")
return
if args.search:
database = UcscDatabase()
found = False
for assembly in database.search(args.search):
found = True
print(f"{assembly.id:12}\t{database.name}\t{assembly.description}")
if not found:
logger.info(f"No match found for {args.search!r}")
return
if args.install:
install_genome(args, config_file)
return
if args.uninstall:
uninstall_genome(args, config_file)
return
def load_installed_pfms(name):
"""Load a pre-installed motif PFMs set.
Parameters
----------
name : str
Name of the pre-installed motif PFMs set to be loaded.
Return
------
pfms : `MotifPfms`
Loaded PFMs of the motif set.
Raises
------
PfmsFileNotFoundError
If the motif PFMs file does not exists.
"""
logger.info(f"Loading motif PFMs set {name!r}")
motif_dir = Config().get_motif_path(name)
pfms_path = pfms_path_fmt.format(motif_dir, name)
if os.path.isfile(pfms_path):
pfms = MotifPfms(name=name)
pfms.read_pfms(path=pfms_path, format='jaspar')
else:
raise PfmsFileNotFoundError(name)
return pfms
def save_built_pwms(self):
"""Save built motif PWMs."""
logger.info(
f"Saving motif PWMs {self.name!r} under assembly {self.genome!r}")
motif_dir = Config().get_motif_path(self.name)
pwms_path = pwms_path_fmt.format(motif_dir, self.name, self.genome)
self.write_motifscan_pwms(pwms_path)
def test_cli_motif_install(motif_root, tmp_dir):
config_file = os.path.join(tmp_dir, "test_cli_motif.motifscanrc")
config = Config(config_file)
config.set_motif_dir(tmp_dir)
config.write()
pfms_path = os.path.join(motif_root, "test", "test_pfms.jaspar")
args = parser.parse_args(
["motif", "--install", "-n", "test_motif", "-i", pfms_path])
run(args=args, config_file=config_file)
motif_path = os.path.join(tmp_dir, "test_motif")
assert os.path.isfile(os.path.join(motif_path, "test_motif_pfms.jaspar"))
config = Config(config_file)
assert config.has_motif_set("test_motif")
assert config.get_motif_path("test_motif") == motif_path
def test_cli_config_set_default_motif(tmp_dir):
args = parser.parse_args(
["config", "--set-default-motif", "/path/to/motif/root"])
config_file = os.path.join(tmp_dir, "test_cli_config.motifscanrc")
run(args=args, config_file=config_file)
config = Config(config_file)
assert config.get_motif_dir() == "/path/to/motif/root"
def test_cli_config_set_motif(tmp_dir):
args = parser.parse_args(
["config", "--set-motif", "motif_set", "/path/to/motif"])
config_file = os.path.join(tmp_dir, 'test_cli_config.motifscanrc')
run(args=args, config_file=config_file)
config = Config(config_file)
assert config.has_motif_set("motif_set")
assert config.get_motif_path("motif_set") == "/path/to/motif"
def test_cli_config_set_genome(tmp_dir):
args = parser.parse_args(
["config", "--set-genome", "hg19", "/path/to/genome"])
config_file = os.path.join(tmp_dir, 'test_cli_config.motifscanrc')
run(args=args, config_file=config_file)
config = Config(config_file)
assert config.has_genome_assembly("hg19")
assert config.get_genome_path("hg19") == "/path/to/genome"
def test_cli_motif_uninstall(motif_root, tmp_dir):
config_file = os.path.join(tmp_dir, "test_cli_motif.motifscanrc")
config = Config(config_file)
config.set_motif_dir(tmp_dir)
config.write()
args = parser.parse_args(["motif", "--uninstall", "test_motif"])
run(args=args, config_file=config_file)
config = Config(config_file)
assert not config.has_motif_set("test_motif")
motif_path = os.path.join(tmp_dir, "test_motif")
assert not os.path.isdir(motif_path)
args = parser.parse_args(["motif", "--uninstall", "test_motif1"])
with pytest.raises(SystemExit):
run(args=args, config_file=config_file)
def test_cli_genome_uninstall(genome_root, tmp_dir):
config_file = os.path.join(tmp_dir, "test_cli_genome.motifscanrc")
config = Config(config_file)
config.set_genome_dir(tmp_dir)
config.write()
args = parser.parse_args(["genome", "--uninstall", "test_genome"])
run(args=args, config_file=config_file)
config = Config(config_file)
assert not config.has_genome_assembly("test_genome")
genome_path = os.path.join(tmp_dir, "test_genome")
assert not os.path.isdir(genome_path)
args = parser.parse_args(["genome", "--uninstall", "test_genome1"])
with pytest.raises(SystemExit):
run(args=args, config_file=config_file)
def test_config_init(config):
assert len(config._config.sections()) == 3
assert config._config.has_section('motifscan')
assert config._config.has_section('genome')
assert config._config.has_section('motif')
assert config._config.get('motifscan', 'genome_dir') == user_genome_dir
assert config._config.get('motifscan', 'motif_dir') == user_motif_dir
config = Config(path=None)
assert config.path == user_rc_path
def test_cli_motif_list(tmp_dir, capsys):
config_file = os.path.join(tmp_dir, "test_cli_motif.motifscanrc")
config = Config(config_file)
config.set_motif_path("motif_set1", "/path/to/motif1")
config.set_motif_path("motif_set2", "/path/to/motif2")
config.write()
args = parser.parse_args(["motif", "--list"])
run(args=args, config_file=config_file)
captured = capsys.readouterr()
assert captured.out == "motif_set1\nmotif_set2\n"
def test_cli_genome_list(tmp_dir, capsys):
config_file = os.path.join(tmp_dir, "test_cli_genome.motifscanrc")
config = Config(config_file)
config.set_genome_path("hg19", "/path/to/genome1")
config.set_genome_path("hg38", "/path/to/genome2")
config.write()
args = parser.parse_args(["genome", "--list"])
run(args=args, config_file=config_file)
captured = capsys.readouterr()
assert captured.out == "hg19\nhg38\n"
def test_cli_genome_install(genome_root, tmp_dir):
config_file = os.path.join(tmp_dir, "test_cli_genome.motifscanrc")
config = Config(config_file)
config.set_genome_dir(tmp_dir)
config.write()
fasta_path = os.path.join(genome_root, "test", "test.fa")
gene_path = os.path.join(genome_root, "test", "test_gene_annotation.txt")
args = parser.parse_args([
"genome", "--install", "-n", "test_genome", "-i", fasta_path, "-a",
gene_path
])
run(args=args, config_file=config_file)
genome_path = os.path.join(tmp_dir, "test_genome")
genome = Genome(name="test_genome", path=genome_path)
assert genome.fetch_sequence("chr1", 0, 10) == "AaTtCcGgNn"
assert genome.genes
config = Config(config_file)
assert config.has_genome_assembly("test_genome")
assert config.get_genome_path("test_genome") == genome_path
def install_genome(args, config_file=None):
config = Config(config_file)
if config.has_genome_assembly(args.name):
logger.error(f"Genome assembly {args.name!r} already exists!")
sys.exit(1)
genome_dir = os.path.abspath(
args.output_dir or os.path.join(config.get_genome_dir(), args.name))
logger.info(f"Installing genome assembly {args.name!r} into {genome_dir}")
if not os.path.isdir(genome_dir):
os.makedirs(genome_dir)
if os.listdir(genome_dir):
logger.error("Directory not empty! Please specify another directory "
"or delete files under it.")
sys.exit(1)
fasta_path = fasta_path_fmt.format(genome_dir, args.name)
bg_freq_path = bg_freq_path_fmt.format(genome_dir, args.name)
gene_path = gene_path_fmt.format(genome_dir, args.name)
if args.remote:
download_dir = os.path.join(genome_dir, 'downloads')
try:
db = UcscDatabase()
dst_fasta = db.download_sequence(args.remote, download_dir)
logger.debug(f"Extracting the sequence file to {fasta_path}")
merge_extracted_files(dst_fasta, fasta_path)
dst_gene = db.download_gene(args.remote, download_dir)
logger.debug(f"Extracting the gene annotation file to {gene_path}")
merge_extracted_files(dst_gene, gene_path)
if args.clean:
logger.debug(f"Removing the download directory {download_dir}")
shutil.rmtree(download_dir)
except RemoteGenomeNotFoundError as e:
logger.error(e)
sys.exit(1)
else:
logger.info("Copying the sequence file(s)")
merge_files(args.fasta_files, fasta_path)
logger.info("Copying the gene annotation file")
copy_file(args.gene_file, gene_path)
logger.info("Calculating nucleotide frequencies of the genome background")
bg_freq = cal_bg_freq(fasta_path)
logger.info("Writing nucleotide frequencies")
write_bg_freq(bg_freq_path, bg_freq)
logger.info("Updating the config file")
config.set_genome_path(args.name, genome_dir)
config.write()
logger.info("Successfully installed!")
def test_cli_config_get_motif(tmp_dir, capsys):
config_file = os.path.join(tmp_dir, "test_cli_config.motifscanrc")
config = Config(config_file)
config.set_motif_path("motif_set", "/path/to/motif")
config.write()
args = parser.parse_args(["config", "--get-motif", "motif_set"])
run(args=args, config_file=config_file)
captured = capsys.readouterr()
assert captured.out == "/path/to/motif\n"
with pytest.raises(SystemExit):
args = parser.parse_args(["config", "--get-motif", "motif_set1"])
run(args=args, config_file=config_file)
def test_cli_config_get_genome(tmp_dir, capsys):
config_file = os.path.join(tmp_dir, "test_cli_config.motifscanrc")
config = Config(config_file)
config.set_genome_path("hg19", "/path/to/genome")
config.write()
args = parser.parse_args(["config", "--get-genome", "hg19"])
run(args=args, config_file=config_file)
captured = capsys.readouterr()
assert captured.out == "/path/to/genome\n"
with pytest.raises(SystemExit):
args = parser.parse_args(["config", "--get-genome", "hg38"])
run(args=args, config_file=config_file)
def test_cli_config_show(tmp_dir, capsys):
config_file = os.path.join(tmp_dir, "test_cli_config.motifscanrc")
config = Config(config_file)
config.set_genome_dir("/path/to/genome/root")
config.set_motif_dir("/path/to/motif/root")
config.set_genome_path("hg19", "/path/to/genome")
config.set_motif_path("motif_set", "/path/to/motif")
config.write()
args = parser.parse_args(["config", "--show"])
run(args=args, config_file=config_file)
captured = capsys.readouterr()
assert captured.out == "[motifscan]\n" \
"genome_dir: /path/to/genome/root\n" \
"motif_dir: /path/to/motif/root\n\n" \
"[genome]\n" \
"hg19: /path/to/genome\n\n" \
"[motif]\n" \
"motif_set: /path/to/motif\n"
def install_motif(args, config_file=None):
config = Config(config_file)
if config.has_motif_set(args.name):
logger.error(f"Motif set {args.name!r} already exists!")
sys.exit(1)
motif_dir = os.path.abspath(
args.output_dir or os.path.join(config.get_motif_dir(), args.name))
logger.info(f"Installing motif set {args.name!r} into {motif_dir}")
if not os.path.isdir(motif_dir):
os.makedirs(motif_dir)
if os.listdir(motif_dir):
logger.error("Directory not empty! Please specify another directory "
"or delete files under it.")
sys.exit(1)
pfms_path = pfms_path_fmt.format(motif_dir, args.name)
if args.remote:
try:
db = JasparDatabase()
if args.database == 'jaspar_core':
dst_pfms = db.download_core(args.remote, motif_dir)
else:
dst_pfms = db.download_other_collections(
args.remote, motif_dir)
logger.debug(
f"Renaming downloaded file to {os.path.basename(pfms_path)}")
shutil.move(dst_pfms, pfms_path)
except RemoteMotifPFMsNotFoundError as e:
logger.error(e)
sys.exit(1)
else:
logger.info("Copying the PFMs file(s)")
merge_files(args.pfm_files, pfms_path)
logger.info("Updating the config file")
config.set_motif_path(args.name, motif_dir)
config.write()
logger.info("Successfully installed!")
if args.genome:
build_motif(args, config_file)
def uninstall_motif(args, config_file=None):
try:
config = Config(config_file)
path = config.get_motif_path(args.uninstall)
except MotifSetNotFoundError as e:
logger.error(e)
sys.exit(1)
logger.info(f"Uninstalling motif set {args.uninstall!r}")
if os.path.isdir(path):
logger.info(f"Removing files under {path}")
try:
shutil.rmtree(path)
except Exception as e:
logger.error(f"Failed to remove the motif directory: {e}")
sys.exit(1)
logger.info("Updating the config file")
config.remove_motif_path(args.uninstall)
config.write()
logger.info("Successfully uninstalled!")
def __init__(self, name, path=None):
logger.info(f"Loading genome {name!r}")
self.name = name
self.path = path or Config().get_genome_path(self.name)
self._fasta_path = fasta_path_fmt.format(self.path, self.name)
self._bg_freq_path = bg_freq_path_fmt.format(self.path, self.name)
self._gene_path = gene_path_fmt.format(self.path, self.name)
if os.path.isfile(self._fasta_path):
self.fa = pysam.FastaFile(self._fasta_path)
else:
raise GenomeFileNotFoundError(self.name, 'sequence')
if os.path.isfile(self._bg_freq_path):
self.bg_freq = read_bg_freq(self._bg_freq_path)
else:
raise GenomeFileNotFoundError(self.name, 'background frequency')
if os.path.isfile(self._gene_path):
self.genes = read_gene_annotation(self._gene_path)
else:
logger.warning("No gene annotation file found")
self.genes = None
self._chroms = None
self._chrom_sizes = None
def load_built_pwms(name, genome):
"""Load built motif PWMs.
Parameters
----------
name : str
Name of the built motif PWMs set to be loaded.
genome : str
Genome assembly name under which these PWMs are built.
Raises
------
PwmsFileNotFoundError
If the motif PWMs file does not exists
"""
logger.info(f"Loading motif PWMs set {name!r} under genome {genome!r}")
motif_dir = Config().get_motif_path(name)
pwms_path = pwms_path_fmt.format(motif_dir, name, genome)
pwms = MotifPwms(name=name, genome=genome)
if os.path.isfile(pwms_path):
pwms.read_motifscan_pwms(pwms_path)
else:
raise PwmsFileNotFoundError(name, genome)
return pwms
def config(tmp_dir):
"""Returns a config instance to test the configuration of MotifScan."""
return Config(os.path.join(tmp_dir, '.motifscanrc'))
def run(args, config_file=None):
config = Config(config_file)
modified = False
if args.show:
print("[motifscan]")
print(f"genome_dir: {config.get_genome_dir()}")
print(f"motif_dir: {config.get_motif_dir()}")
print("\n[genome]")
for name, path in config.list_genome_assemblies():
print(f"{name}: {path}")
print("\n[motif]")
for name, path in config.list_motif_sets():
print(f"{name}: {path}")
return
if args.set_default_genome:
logger.debug(
"Setting the default installation path for genome assemblies")
config.set_genome_dir(os.path.abspath(args.set_default_genome))
modified = True
if args.set_default_motif:
logger.debug("Setting the default installation path for motif sets")
config.set_motif_dir(os.path.abspath(args.set_default_motif))
modified = True
if args.get_genome:
logger.debug(f"Getting the genome path of {args.get_genome!r}")
try:
print(config.get_genome_path(args.get_genome))
except GenomeNotFoundError as e:
logger.error(e)
sys.exit(1)
if args.set_genome:
name = args.set_genome[0]
path = os.path.abspath(args.set_genome[1])
logger.debug(f"Setting the genome path for {name!r}: {path}")
config.set_genome_path(name, path)
modified = True
if args.rm_genome:
logger.debug(f"Removing the genome path for {args.rm_genome!r}")
try:
config.remove_genome_path(args.rm_genome)
modified = True
except GenomeNotFoundError as e:
logger.error(e)
sys.exit(1)
if args.get_motif:
logger.debug(f"Getting the motif path of {args.get_motif!r}")
try:
print(config.get_motif_path(args.get_motif))
except MotifSetNotFoundError as e:
logger.error(e)
sys.exit(1)
if args.set_motif:
name = args.set_motif[0]
path = os.path.abspath(args.set_motif[1])
logger.debug(f"Setting the motif path for {name!r}: {path}")
config.set_motif_path(name, path)
modified = True
if args.rm_motif:
logger.debug(f"Removing the motif path for {args.rm_motif!r}")
try:
config.remove_motif_path(args.rm_motif)
modified = True
except MotifSetNotFoundError as e:
logger.error(e)
sys.exit(1)
if modified:
logger.debug(f"Updating the config file: {config.path}")
config.write()
logger.debug("Done")
def test_invalid_config(data_dir):
with pytest.raises(InvalidConfigFileError):
Config(os.path.join(data_dir, 'invalid.motifscanrc'))