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)
