def ensembl(args):
"""
%prog ensembl species
Retrieve genomes and annotations from ensembl FTP. Available species
listed below. Use comma to give a list of species to download. For example:
$ %prog ensembl danio_rerio,gasterosteus_aculeatus
"""
p = OptionParser(ensembl.__doc__)
p.add_option("--version", default="75", help="Ensembl version")
opts, args = p.parse_args(args)
version = opts.version
url = "ftp://ftp.ensembl.org/pub/release-{0}/".format(version)
fasta_url = url + "fasta/"
valid_species = [x for x in ls_ftp(fasta_url) if "." not in x]
doc = "\n".join((ensembl.__doc__, tile(valid_species)))
p.set_usage(doc)
if len(args) != 1:
sys.exit(not p.print_help())
(species,) = args
species = species.split(",")
for s in species:
download_species_ensembl(s, valid_species, url)
def phytozome(args):
"""
%prog phytozome species
Retrieve genomes and annotations from phytozome FTP. Available species
listed below. Use comma to give a list of species to download. For example:
$ %prog phytozome Athaliana,Vvinifera,Osativa,Sbicolor,Slycopersicum
"""
p = OptionParser(phytozome.__doc__)
p.add_option("--version", default="9.0",
help="Phytozome version [default: %default]")
p.add_option("--assembly", default=False, action="store_true",
help="Download assembly [default: %default]")
opts, args = p.parse_args(args)
url = "ftp://ftp.jgi-psf.org/pub/compgen/phytozome/v{0}/".\
format(opts.version)
valid_species = [x for x in ls_ftp(url) if "." not in x]
doc = "\n".join((phytozome.__doc__, tile(valid_species)))
p.set_usage(doc)
if len(args) != 1:
sys.exit(not p.print_help())
species, = args
species = species.split(",")
for s in species:
download_species_phytozome(s, valid_species, url, assembly=opts.assembly)
def ensembl(args):
"""
%prog ensembl species
Retrieve genomes and annotations from ensembl FTP. Available species
listed below. Use comma to give a list of species to download. For example:
$ %prog ensembl danio_rerio,gasterosteus_aculeatus
"""
p = OptionParser(ensembl.__doc__)
p.add_option("--version", default="75",
help="Ensembl version [default: %default]")
opts, args = p.parse_args(args)
version = opts.version
url = "ftp://ftp.ensembl.org/pub/release-{0}/".format(version)
fasta_url = url + "fasta/"
valid_species = [x for x in ls_ftp(fasta_url) if "." not in x]
doc = "\n".join((ensembl.__doc__, tile(valid_species)))
p.set_usage(doc)
if len(args) != 1:
sys.exit(not p.print_help())
species, = args
species = species.split(",")
for s in species:
download_species_ensembl(s, valid_species, url)
def phytozome(args):
"""
%prog phytozome species
Retrieve genomes and annotations from phytozome FTP. Available species
listed below. Use comma to give a list of species to download. For example:
$ %prog phytozome Athaliana,Vvinifera,Osativa,Sbicolor,Slycopersicum
"""
from jcvi.formats.gff import bed as gff_bed
from jcvi.formats.fasta import format as fasta_format
p = OptionParser(phytozome.__doc__)
p.add_option("--version", default="9.0",
help="Phytozome version [default: %default]")
p.add_option("--assembly", default=False, action="store_true",
help="Download assembly [default: %default]")
p.add_option("--format", default=False, action="store_true",
help="Format to CDS and BED for synteny inference")
opts, args = p.parse_args(args)
url = "ftp://ftp.jgi-psf.org/pub/compgen/phytozome/v{0}/".\
format(opts.version)
valid_species = [x for x in ls_ftp(url) if "." not in x]
doc = "\n".join((phytozome.__doc__, tile(valid_species)))
p.set_usage(doc)
if len(args) != 1:
sys.exit(not p.print_help())
species, = args
if species == "all":
species = ",".join(valid_species)
species = species.split(",")
use_IDs = set()
# We have to watch out when the gene names and mRNA names mismatch, in which
# case we just extract the mRNA names
use_mRNAs = set(["Cclementina", "Creinhardtii", "Csinensis", "Fvesca",
"Lusitatissimum", "Mesculenta", "Mguttatus", "Ppersica",
"Pvirgatum", "Rcommunis", "Sitalica", "Tcacao",
"Thalophila", "Vcarteri", "Vvinifera", "Zmays"])
for s in species:
gff, fa = download_species_phytozome(s, valid_species, url,
assembly=opts.assembly)
key = "ID" if s in use_IDs else "Name"
ttype = "mRNA" if s in use_mRNAs else "gene"
if not opts.format:
continue
bedfile = s + ".bed"
cdsfile = s + ".cds"
gff_bed([gff, "--type={}".format(ttype), "--key={}".format(key),
"-o", bedfile])
fasta_format([fa, cdsfile, r"--sep=|"])
def phytozome(args):
"""
%prog phytozome species
Retrieve genomes and annotations from phytozome FTP. Available species
listed below. Use comma to give a list of species to download. For example:
$ %prog phytozome Athaliana,Vvinifera,Osativa,Sbicolor,Slycopersicum
"""
from jcvi.formats.gff import bed as gff_bed
from jcvi.formats.fasta import format as fasta_format
p = OptionParser(phytozome.__doc__)
p.add_option("--version", default="9.0",
help="Phytozome version [default: %default]")
p.add_option("--assembly", default=False, action="store_true",
help="Download assembly [default: %default]")
p.add_option("--format", default=False, action="store_true",
help="Format to CDS and BED for synteny inference")
opts, args = p.parse_args(args)
url = "ftp://ftp.jgi-psf.org/pub/compgen/phytozome/v{0}/".\
format(opts.version)
valid_species = [x for x in ls_ftp(url) if "." not in x]
doc = "\n".join((phytozome.__doc__, tile(valid_species)))
p.set_usage(doc)
if len(args) != 1:
sys.exit(not p.print_help())
species, = args
if species == "all":
species = ",".join(valid_species)
species = species.split(",")
use_IDs = set()
# We have to watch out when the gene names and mRNA names mismatch, in which
# case we just extract the mRNA names
use_mRNAs = set(["Cclementina", "Creinhardtii", "Csinensis", "Fvesca",
"Lusitatissimum", "Mesculenta", "Mguttatus", "Ppersica",
"Pvirgatum", "Rcommunis", "Sitalica", "Tcacao",
"Thalophila", "Vcarteri", "Vvinifera", "Zmays"])
for s in species:
gff, fa = download_species_phytozome(s, valid_species, url,
assembly=opts.assembly)
key = "ID" if s in use_IDs else "Name"
ttype = "mRNA" if s in use_mRNAs else "gene"
if not opts.format:
continue
bedfile = s + ".bed"
cdsfile = s + ".cds"
gff_bed([gff, "--type={}".format(ttype), "--key={}".format(key),
"-o", bedfile])
fasta_format([fa, cdsfile, r"--sep=|"])
def phytozome9(args):
"""
%prog phytozome9 species
Retrieve genomes and annotations from phytozome FTP. Available species
listed below. Use comma to give a list of species to download. For example:
$ %prog phytozome9 Athaliana,Vvinifera,Osativa,Sbicolor,Slycopersicum
"""
p = OptionParser(phytozome9.__doc__)
p.add_option(
"--assembly",
default=False,
action="store_true",
help="Download assembly",
)
p.add_option(
"--format",
default=False,
action="store_true",
help="Format to CDS and BED for synteny inference",
)
opts, args = p.parse_args(args)
version = "9.0"
url = "ftp://ftp.jgi-psf.org/pub/compgen/phytozome/v{0}/".format(version)
valid_species = [x for x in ls_ftp(url) if "." not in x]
doc = "\n".join((phytozome9.__doc__, tile(valid_species)))
p.set_usage(doc)
if len(args) != 1:
sys.exit(not p.print_help())
(species, ) = args
if species == "all":
species = ",".join(valid_species)
species = species.split(",")
for s in species:
res = download_species_phytozome9(s,
valid_species,
url,
assembly=opts.assembly)
if not res:
logging.error("No files downloaded")
gff, cdsfa = res.get("gff"), res.get("cds")
if opts.format:
format_bed_and_cds(s, gff, cdsfa)
def phytozome(args):
"""
%prog phytozome species
Retrieve genomes and annotations from phytozome FTP. Available species
listed below. Use comma to give a list of species to download. For example:
$ %prog phytozome Athaliana,Vvinifera,Osativa,Sbicolor,Slycopersicum
"""
from jcvi.formats.gff import bed as gff_bed
from jcvi.formats.fasta import format as fasta_format
p = OptionParser(phytozome.__doc__)
p.add_option("--version",
default="9.0",
help="Phytozome version [default: %default]")
p.add_option("--assembly",
default=False,
action="store_true",
help="Download assembly [default: %default]")
p.add_option("--format",
default=False,
action="store_true",
help="Format to CDS and BED for synteny inference")
opts, args = p.parse_args(args)
url = "ftp://ftp.jgi-psf.org/pub/compgen/phytozome/v{0}/".\
format(opts.version)
valid_species = [x for x in ls_ftp(url) if "." not in x]
doc = "\n".join((phytozome.__doc__, tile(valid_species)))
p.set_usage(doc)
if len(args) != 1:
sys.exit(not p.print_help())
species, = args
species = species.split(",")
for s in species:
gff, fa = download_species_phytozome(s,
valid_species,
url,
assembly=opts.assembly)
if not opts.format:
continue
bedfile = s + ".bed"
cdsfile = s + ".cds"
gff_bed([gff, "--phytozome", "-o", bedfile])
fasta_format([fa, cdsfile, r"--sep=|"])
def phytozome(args):
"""
%prog phytozome species
Retrieve genomes and annotations from phytozome FTP. Available species
listed below. Use comma to give a list of species to download. For example:
$ %prog phytozome Athaliana,Vvinifera,Osativa,Sbicolor,Slycopersicum
"""
p = OptionParser(phytozome.__doc__)
p.add_option("--version",
default="9.0",
help="Phytozome version [default: %default]")
p.add_option("--assembly",
default=False,
action="store_true",
help="Download assembly [default: %default]")
opts, args = p.parse_args(args)
url = "ftp://ftp.jgi-psf.org/pub/compgen/phytozome/v{0}/".\
format(opts.version)
valid_species = [x for x in ls_ftp(url) if "." not in x]
doc = "\n".join((phytozome.__doc__, tile(valid_species)))
p.set_usage(doc)
if len(args) != 1:
sys.exit(not p.print_help())
species, = args
species = species.split(",")
for s in species:
download_species_phytozome(s,
valid_species,
url,
assembly=opts.assembly)
def phytozome(args):
"""
%prog phytozome species
Retrieve genomes and annotations from phytozome using Globus API. Available
species listed below. Use comma to give a list of species to download. For
example:
$ %prog phytozome Athaliana,Vvinifera,Osativa,Sbicolor,Slycopersicum
The downloader will prompt you to enter Phytozome user name and password
during downloading. Please register for a login at:
https://phytozome.jgi.doe.gov/pz/portal.html.
"""
from jcvi.apps.biomart import GlobusXMLParser
p = OptionParser(phytozome.__doc__)
p.add_option(
"--version",
default="12",
choices=("9", "10", "11", "12", "12_unrestricted", "13"),
help="Phytozome version",
)
p.add_option(
"--assembly",
default=False,
action="store_true",
help="Download assembly",
)
p.add_option(
"--format",
default=False,
action="store_true",
help="Format to CDS and BED for synteny inference",
)
p.set_downloader()
opts, args = p.parse_args(args)
downloader = opts.downloader
directory_listing = ".phytozome_directory_V{}.xml".format(opts.version)
# Get directory listing
base_url = "http://genome.jgi.doe.gov"
dlist = "{}/ext-api/downloads/get-directory?organism=PhytozomeV{}".format(
base_url, opts.version
)
# Make sure we have a valid cookies
cookies = get_cookies()
if cookies is None:
logging.error("Error fetching cookies ... cleaning up")
FileShredder([directory_listing])
sys.exit(1)
# Proceed to use the cookies and download the species list
try:
download(
dlist,
filename=directory_listing,
cookies=cookies,
downloader=downloader,
)
g = GlobusXMLParser(directory_listing)
except:
logging.error("Error downloading directory listing ... cleaning up")
FileShredder([directory_listing, cookies])
sys.exit(1)
genomes = g.get_genomes()
valid_species = genomes.keys()
species_tile = tile(valid_species)
p.set_usage("\n".join((phytozome.__doc__, species_tile)))
if len(args) != 1:
sys.exit(not p.print_help())
(species,) = args
if species == "all":
species = ",".join(valid_species)
species = species.split(",")
for s in species:
res = download_species_phytozome(
genomes,
s,
valid_species,
base_url,
cookies,
assembly=opts.assembly,
downloader=downloader,
)
if not res:
logging.error("No files downloaded")
gff, fa = res.get("gff"), res.get("cds")
if opts.format:
format_bed_and_cds(s, gff, fa)
