import random
from Bio.PDB import PDBList
try:
DATASIZE = int(sys.argv[1])
except:
print('Usage:', os.path.basename(sys.argv[0]), '''N
N - number of entries to be randomly selected from all PDB entries.
Output written to stdout.''', file=sys.stderr)
sys.exit(1)
# Get index file
pdbl = PDBList()
#print(f"Downloading PDB index file, this takes a while.", file=sys.stderr)
all_entries = pdbl.get_all_entries()
# Select random subset
selected = random.sample(all_entries, DATASIZE)
print(f"Randomly selected %d entries from {len(all_entries)} entries." % DATASIZE, file=sys.stderr)
# Write results to stdout
for entry in selected:
print(entry, file=sys.stdout)
# Get data files
#for entry in selected:
# pdbl.retrieve_pdb_file(entry, file_format="pdb", pdir="pdb")
# Test which PDB entries error on PDB/mmCIF parsers
# Writes output to a file labelled with the week
import os
from datetime import datetime
from math import ceil
from Bio.PDB import PDBList
from Bio.PDB.PDBParser import PDBParser
from Bio.PDB.MMCIFParser import MMCIFParser
start = datetime.now()
basedir = "."
pdbl = PDBList()
pdblist = pdbl.get_all_entries()
outstrs = [
"Checking all PDB entries at {}".format(start.isoformat()),
"Checking {} entries".format(len(pdblist))
]
pdb_parser = PDBParser()
mmcif_parser = MMCIFParser()
for pu in sorted(pdblist):
p = pu.lower()
try:
pdbl.retrieve_pdb_file(p, pdir=basedir, file_format="pdb")
except:
# Not having a PDB file is acceptable, though a failure to download an
# available file may hide an error in parsing
try:
from configure import configure
config = configure()
from Bio.PDB import PDBList
from os.path import dirname, join, basename
from glob import glob
from os import remove
pdb_dir = dirname(config["cif"])
pl = PDBList(pdb=pdb_dir)
pl.flat_tree = True
existingCifs = glob(join(pdb_dir, "*.cif"))
existingPdbs = set([])
for cif in existingCifs:
existingPdbs.add(basename(cif).replace(".cif", ""))
allPdbs = set(pdb_code.lower() for pdb_code in pl.get_all_entries())
pdb2delete = existingPdbs - allPdbs
pdb2download = allPdbs - existingPdbs
print("Found ", len(pdb2delete), " files to delete")
print("and ", len(pdb2download), " file to download")
for pdb_code in pdb2download:
pl.retrieve_pdb_file(pdb_code, file_format="mmCif")
for pdb_code in pdb2delete:
cifPath = join(pdb_dir, pdb_code + ".cif")
remove(cifPath)
