def get_genome_download_link(self, name, mask="soft", **kwargs):
"""
Return NCBI ftp link to top-level genome sequence
Parameters
----------
name : str
Genome name. Current implementation will fail if exact
name is not found.
Returns
------
tuple (name, link) where name is the NCBI asm_name identifier
and link is a str with the ftp download link.
"""
if mask != "soft":
sys.stderr.write("ignoring mask parameter for NCBI at download.\n")
if not self.genomes:
self.genomes = self._get_genomes()
for genome in self.genomes:
if name in [genome["asm_name"], genome["asm_name"].replace(" ", "_")]:
url = genome["ftp_path"]
url = url.replace("ftp://", "https://")
url += "/" + url.split("/")[-1] + "_genomic.fna.gz"
return name, url
raise exceptions.GenomeDownloadError("Could not download genome from NCBI")
def get_genome_download_link(self, name, mask="soft", version=None):
"""
Return NCBI ftp link to top-level genome sequence
Parameters
----------
name : str
Genome name. Current implementation will fail if exact
name is not found.
Returns
------
tuple (name, link) where name is the NCBI asm_name identifier
and link is a str with the ftp download link.
"""
if mask == "hard":
sys.stderr.write("ignoring mask parameter for NCBI at download.\n")
if not self.genomes:
self.genomes = self._get_genomes()
for genome in self.genomes:
if genome["asm_name"] == name:
#ftp_path': 'ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/004/195/GCF_000004195.3_Xenopus_tropicalis_v9.1'
url = genome["ftp_path"]
url += "/" + url.split("/")[-1] + "_genomic.fna.gz"
return name, url
raise exceptions.GenomeDownloadError(
"Could not download genome from NCBI")
def get_genome_download_link(self, name, mask="soft", **kwargs):
"""
Return UCSC http link to genome sequence
Parameters
----------
name : str
Genome build name. Current implementation will fail if exact
name is not found.
Returns
------
tuple (name, link) where name is the genome build identifier
and link is a str with the http download link.
"""
if mask not in ["soft", "hard"]:
sys.stderr.write("ignoring mask parameter for UCSC at download.\n")
urls = [self.ucsc_url, self.alt_ucsc_url]
if mask == "hard":
urls = [self.ucsc_url_masked, self.alt_ucsc_url_masked]
for genome_url in urls:
remote = genome_url.format(name)
ret = requests.head(remote)
if ret.status_code == 200:
return name, remote
raise exceptions.GenomeDownloadError(
"Could not download genome {} from UCSC".format(name)
)
def _get_genome_info(self, name):
"""Get genome_info from json request."""
try:
assembly_id = ""
for genome in self.list_available_genomes(as_dict=True):
if genome.get("assembly_name", "") == name:
assembly_id = genome.get("assembly_id", "")
break
if assembly_id:
ext = "info/genomes/assembly/" + assembly_id + "/?"
genome_info = self.request_json(ext)
else:
raise exceptions.GenomeDownloadError(
"Could not download genome {} from Ensembl".format(name))
except requests.exceptions.HTTPError as e:
sys.stderr.write("Species not found: {}".format(e))
raise exceptions.GenomeDownloadError(
"Could not download genome {} from Ensembl".format(name))
return genome_info
