def download_gencode_gff(gff_url, output_dir, decompress):
"""
Download GENCODE GFF3 file.
Updated 2020-02-09.
"""
download(
url=gff_url,
output_dir=os.path.join(output_dir, "gff"),
decompress=decompress,
)
def download_refseq_gtf(build, release_url, output_dir, decompress):
"""
Download RefSeq GTF file.
Updated 2020-02-09.
"""
download(
url=paste_url(release_url, build + "_latest_genomic.gtf.gz"),
output_dir=os.path.join(output_dir, "gtf"),
decompress=decompress,
)
def download_gencode_genome(genome_fasta_url, output_dir, decompress):
"""
Download GENCODE genome FASTA.
Updated 2020-02-09.
"""
download(
url=genome_fasta_url,
output_dir=os.path.join(output_dir, "genome"),
decompress=decompress,
)
def download_refseq_transcriptome(build, release_url, output_dir, decompress):
"""
Download RefSeq transcriptome FASTA.
Updated 2020-02-09.
"""
download(
url=paste_url(release_url, build + "_latest_rna.fna.gz"),
output_dir=os.path.join(output_dir, "transcriptome"),
decompress=decompress,
)
def download_flybase_gff(release_url, output_dir, decompress, dmel):
"""
Download FlyBase GFF3 file.
Updated 2020-02-09.
"""
output_dir = os.path.join(output_dir, "gff")
gff_url = paste_url(release_url, "gff")
download(url=paste_url(gff_url, "md5sum.txt"), output_dir=output_dir)
download(
url=paste_url(gff_url, "dmel-all-" + dmel + ".gff.gz"),
output_dir=output_dir,
decompress=decompress,
)
def download_flybase_genome(release_url, output_dir, decompress, dmel):
"""
Download genome FASTA.
Updated 2020-02-09.
"""
output_dir = os.path.join(output_dir, "genome")
fasta_url = paste_url(release_url, "fasta")
download(url=paste_url(fasta_url, "md5sum.txt"), output_dir=output_dir)
download(
url=paste_url(fasta_url, "dmel-all-aligned-" + dmel + ".fasta.gz"),
output_dir=output_dir,
decompress=decompress,
)
def download_ensembl_transcriptome(organism, build, release_url, output_dir,
decompress):
"""
Download Ensembl transcriptome FASTA.
Updated 2020-02-09.
"""
output_dir = os.path.join(output_dir, "transcriptome")
base_url = paste_url(release_url, "fasta", organism.lower(), "cdna")
readme_url = paste_url(base_url, "README")
checksums_url = paste_url(base_url, "CHECKSUMS")
fasta_url = paste_url(base_url, organism + "." + build + ".cdna.all.fa.gz")
download(url=readme_url, output_dir=output_dir)
download(url=checksums_url, output_dir=output_dir)
download(url=fasta_url, output_dir=output_dir, decompress=decompress)
def download_ensembl_gff(organism, build, release, release_url, output_dir,
decompress):
"""
Download Ensembl GFF3 file.
Updated 2020-02-09.
"""
output_dir = os.path.join(output_dir, "gff")
base_url = paste_url(release_url, "gff3", organism.lower())
readme_url = paste_url(base_url, "README")
checksums_url = paste_url(base_url, "CHECKSUMS")
gff_url = paste_url(base_url,
organism + "." + build + "." + release + ".gff3.gz")
download(url=readme_url, output_dir=output_dir)
download(url=checksums_url, output_dir=output_dir)
download(url=gff_url, output_dir=output_dir, decompress=decompress)
if organism in ("H**o sapiens", "Mus musculus"):
gtf_patch_url = paste_url(
base_url,
organism + "." + build + "." + release +
".chr_patch_hapl_scaff.gff3.gz",
)
download(url=gtf_patch_url,
output_dir=output_dir,
decompress=decompress)
def download_ensembl_genome(organism, build, release_url, output_dir,
decompress):
"""
Download Ensembl genome FASTA.
Updated 2020-02-09.
"""
output_dir = os.path.join(output_dir, "genome")
base_url = paste_url(release_url, "fasta", organism.lower(), "dna")
readme_url = paste_url(base_url, "README")
checksums_url = paste_url(base_url, "CHECKSUMS")
if organism in ("Homo_sapiens", "Mus_musculus"):
assembly = "primary_assembly"
else:
assembly = "toplevel"
fasta_url = paste_url(
base_url, organism + "." + build + ".dna." + assembly + ".fa.gz")
download(url=readme_url, output_dir=output_dir)
download(url=checksums_url, output_dir=output_dir)
download(url=fasta_url, output_dir=output_dir, decompress=decompress)
def download_flybase_transcriptome(release_url, output_dir, decompress, dmel):
"""
Download FlyBase transcriptome FASTA.
Updated 2020-02-09.
"""
output_dir = os.path.join(output_dir, "transcriptome")
cat_dir = os.path.join(output_dir, "cat")
output_fasta_file = os.path.join(
output_dir, "dmel-transcriptome-" + dmel + ".fasta.gz")
fasta_url = paste_url(release_url, "fasta")
download(url=paste_url(fasta_url, "md5sum.txt"), output_dir=cat_dir)
download(
url=paste_url(fasta_url, "dmel-all-transcript-" + dmel + ".fasta.gz"),
output_dir=cat_dir,
)
download(
url=paste_url(fasta_url, "dmel-all-miRNA-" + dmel + ".fasta.gz"),
output_dir=cat_dir,
)
download(
url=paste_url(fasta_url, "dmel-all-miscRNA-" + dmel + ".fasta.gz"),
output_dir=cat_dir,
)
download(
url=paste_url(fasta_url, "dmel-all-ncRNA-" + dmel + ".fasta.gz"),
output_dir=cat_dir,
)
download(
url=paste_url(fasta_url, "dmel-all-pseudogene-" + dmel + ".fasta.gz"),
output_dir=cat_dir,
)
download(
url=paste_url(fasta_url, "dmel-all-tRNA-" + dmel + ".fasta.gz"),
output_dir=cat_dir,
)
if not os.path.isfile(output_fasta_file):
print("Concatenating '" + output_fasta_file + "'.")
fasta_glob = os.path.join(cat_dir, "dmel-all-*.fasta.gz")
shell("cat " + fasta_glob + " > " + output_fasta_file)
if decompress is True:
decompress_but_keep_original(output_fasta_file)