def parsepdb(iqmain, iqart): #general pipeline, tentative
"""fetch IDs from pdb that contain zinc, returns two list of IDs, one for zinc and one for zinc x-ray artifacts"""
qmain = pd.make_query(iqmain, querytype='AdvancedKeywordQuery')
qart = pd.make_query(iqart, querytype='AdvancedKeywordQuery')
qmainl = []
qartfinal = []
for hit in pd.do_search(qmain):
qmainl.append(hit)
for hit in pd.do_search(qart):
qartfinal.append(hit)
return qmainl, qartfinal
def pdblistfilter(qmainl, iqnmr, iqxray):
"""filters out zinc hits for NMR or x-ray methodology"""
qnmr = pd.make_query(iqnmr, querytype='ExpTypeQuery')
qxray = pd.make_query(iqxray, querytype='ExpTypeQuery')
qnmrl = []
qxrayl = []
for hit in pd.do_search(qnmr):
qnmrl.append(hit)
for hit in pd.do_search(qxray):
qxrayl.append(hit)
qnmrfinal = []
qxrayfinal = []
for element in qmainl:
if element in qnmrl:
qnmrfinal.append(element)
if element in qxrayl:
qxrayfinal.append(element)
return qnmrfinal, qxrayfinal
def searchPDB(searchTerm):
# use these entries for test purposes
#pdbtest = '3vtv' #BUG: this structure uses entityNr instead of nr_entities
pdbtest = '4xkl'
#pdbtest = '5v4k'
searchStructures = make_query(pdbtest, querytype='AdvancedKeywordQuery')
#searchStructures = make_query(searchTerm, querytype='AdvancedKeywordQuery')
foundStructures = do_search(searchStructures)
print(foundStructures)
return (foundStructures)
def find_matches(query, df):
"""
Search the PDB database for matches to a given query using pypdb,
then cross-reference the results with the dataframe provided,
and return a subset of the dataframe with matching pdbid's.
Assumes that the provided which contains a column called 'pdbid'
Returns:
-------
df : pandas.DataFrame
A subset of the provided dataframe, which only includes the
pdbid's which matched the query
"""
# make a PDB database query and perform a search,
# then convert the results to lower case
search_results = [x.lower() for x in pdb.do_search(pdb.make_query(query))]
return df[df['pdbid'].isin(search_results)]
shutil.rmtree(check)
os.makedirs(check)
break
else:
check=input("Please re-write the output directory again: ") # ex: /home/a/Desktop/pdb_data2
os.makedirs(check)
else:
os.makedirs(check)
# Let me change the directory
os.chdir(check)
for each in my_query:
search_dict = make_query(each)
found_pdbs = do_search(search_dict)
if len(found_pdbs)>number:
found_pdbs = found_pdbs[0:number]
for ids in found_pdbs:
# I created all the paths for each pdb ID
ext="{}.pdb".format(ids)
search_path= 'https://files.rcsb.org/view'
full_path=os.path.join(search_path,ext)
# I created a new folder for each gene name in the given list
my_new_folder= os.path.join(check,each)
if not os.path.isdir(my_new_folder):
os.makedirs(my_new_folder)
new_full = os.path.join(my_new_folder,ext)