def human_genome():
import gzip
base = "http://hgdownload.cse.ucsc.edu/goldenpath/hg19/chromosomes/chr{}.fa.gz"
links = {}
for i in range(1, 23):
links[f"chr{i}.dna"] = base.format(i)
links["chrX.dna"] = base.format("X")
links["chrY.dna"] = base.format("Y")
if not os.path.isdir("human"):
os.mkdir("human")
for fn, link in links.items():
print(f"Downloading {fn}...")
data = fetch_bin(link)
print(f"Decompressing {fn}...")
data = gzip.decompress(data)
print(f"Stripping {fn}...")
data = utils.strip_fasta(data.decode())
with open(os.path.join("human", fn), "w") as file:
file.write(data)
def strip_fasta():
infn, outfn = require(2, "<input> <output>")
with open(infn, "r") as file:
content = file.read()
content = utils.strip_fasta(content)
with open(outfn, "w") as file:
file.write(content)
def cat_genome():
links = {"a1": "CM001378.3", "a2": "CM001379.3", "a3": "CM001380.3", "b1": "CM001381.3", "b2": "CM001382.3", "b3": "CM001383.3", "b4": "CM001384.3", "c1": "CM001385.3", "c2": "CM001386.3", "d1": "CM001387.3", "d2": "CM001388.3", "d3": "CM001389.3", "d4": "CM001390.3", "e1": "CM001391.3", "e2": "CM001392.3", "e3": "CM001393.3", "f1": "CM001394.3", "f2": "CM001395.3", "x": "CM001396.3"}
if not os.path.isdir("cat"):
os.mkdir("cat")
for fn, link in links.items():
print(f"Downloading {fn}...")
data = fetch_nuccore(link)
print(f"Stripping {fn}...")
data = utils.strip_fasta(data)
with open(f"cat/chr{fn.upper()}.dna", "w") as file:
file.write(data)
def corona():
links = {"sars_cov2": "NC_045512"}
if not os.path.isdir("corona"):
os.mkdir("corona")
for fn, link in links.items():
print(f"Downloading {fn}...")
data = fetch_nuccore(link)
print(f"Stripping {fn}...")
data = utils.strip_fasta(data)
with open(f"corona/{fn}.dna", "w") as file:
file.write(data)
def dog_genome():
links = {}
ptr = 25100
format = "CM{}.1"
for i in (list(range(1, 39)) + ["X", "Y", "MT"]):
links[str(i)] = format.format(str(ptr).zfill(6))
ptr += 1
if not os.path.isdir("dog"):
os.mkdir("dog")
for fn, link in links.items():
print(f"Downloading {fn}...")
data = fetch_nuccore(link)
print(f"Stripping {fn}...")
data = utils.strip_fasta(data)
with open(f"dog/chr{fn.upper()}.dna", "w") as file:
file.write(data)
def mouse_genome():
links = {}
ptr = 994
format = "CM{}.3"
for i in (list(range(1, 20)) + ["X", "Y"]):
links[str(i)] = format.format(str(ptr).zfill(6))
ptr += 1
links["MT"] = "AY172335.1"
if not os.path.isdir("mouse"):
os.mkdir("mouse")
for fn, link in links.items():
print(f"Downloading {fn}...")
data = fetch_nuccore(link)
print(f"Stripping {fn}...")
data = utils.strip_fasta(data)
with open(f"mouse/chr{fn.upper()}.dna", "w") as file:
file.write(data)