def initWithFlatFile(self, filepath, useGzip=False):
"""
Initialize the database cache with the content of a ott flat file.
"""
if not useGzip:
tOpen = open
else:
tOpen = gzip.open
mapDict = {}
with tOpen(filepath) as flatfile:
flatfile.next()
#skip header
for line in flatfile:
arr = line.strip().split("\t|\t")
if not arr[0] or not arr[4]:
continue
try:
source = dict(
[entry.split(":") for entry in arr[4].split(",")])
except Exception as e:
import pdb
pdb.set_trace()
try:
mapDict[arr[0]] = source["ncbi"]
except KeyError:
pass
self.cacheList[0] = SqliteCache(filePath=self.cacheList[0].filePath,
indict=mapDict,
**self.cacheList[0].conf)
def __init__(self, dbPath):
kegg = KeggKoIdToDefMap(useCache=False)
database = SqliteCache(filePath=dbPath,
indict=None,
table="keggkoId2def",
key="koId",
value="koDef")
MultiCachedDict.__init__(self, None, [database, kegg])
def __init__(self, dbPath):
kegg = KeggPathwayIdToNameMap(useCache=False)
database = SqliteCache(filePath=dbPath,
indict=None,
table="keggpathwayId2name",
key="pathId",
value="pathName")
MultiCachedDict.__init__(self, None, [database, kegg])
def __init__(self, dbPath, email):
ncbi = TaxonomyParentMap(email, useCache=False)
database = SqliteCache(filePath=dbPath,
indict=None,
table="tax2parent",
key="tax",
value="parent")
MultiCachedDict.__init__(self, None, [database, ncbi])
def __init__(self, dbPath, email):
ncbi = NuclId2TaxIdMap(email, useCache=False)
database = SqliteCache(filePath=dbPath,
indict=None,
table="gi2tax",
key="gi",
value="tax")
MultiCachedDict.__init__(self, None, [database, ncbi])
def initWithUniprotFlatFile(filepath, gzip=True):
"""
Initialize the database cache with the content of a uniprot flat file.
This will only work if source and target ID type a part of the
flat file. If the file is not gzipped this has to be specified.
"""
if not gzip:
tOpen = open
else:
tOpen = gzip.open
name2pos = {
"ACC": 0,
"ID": 1,
"P_ENTREZGENEID": 2,
"P_REFSEQ_AC": 3,
"P_GI": 4,
"PDB_ID": 5,
"NF100": 7,
"NF90": 8,
"NF50": 9,
"UPARC": 10,
"PIR": 11,
"MIM_ID": 13,
"UNIGENE_ID": 14,
"EMBL_ID": 16,
"EMBL": 17,
"ENSEMBL_ID": 18,
"ENSEMBL_TRS_ID": 19,
"ENSEMBL_PRO_ID": 20,
}
if self.target not in name2pos:
raise ValueError(
"Map target '%s' is not available from flat file." %
self.target)
targetPos = name2pos[target]
if self.source not in name2pos:
raise ValueError(
"Map source '%s' is not available from flat file." %
self.source)
sourcePos = name2pos[source]
mapDict = {}
for line in tOpen(filepath):
arr = line.strip().split("\t")
if not arr[targetPos] or not arr[sourcePos]:
continue
mapDict[arr[sourcePos]] = arr[targetPos]
self.cacheList[0] = SqliteCache(filePath=self.cacheList[0].filepath,
indict=mapDict,
**self.cacheList[0].conf)
def __init__(self,
dbpath,
context=None,
tablename="ott2ncbi",
keyname="ottId",
valuename="ncbiTaxId"):
dbMap = SqliteCache(dbpath,
table=tablename,
key=keyname,
value=valuename)
ottMap = OttId2NcbiTaxIdMap()
MultiCachedDict.__init__(self, None, [dbMap, ottMap])
def __init__(self,
dbpath,
source="ACC",
target="P_REFSEQ_AC",
retry=0,
delay=1,
contact=None,
returnNone=False,
tablename=None,
keyname=None,
valuename=None):
self.source = source
self.target = target
dbMap = SqliteCache(dbpath,
table=tablename,
key=keyname,
value=valuename)
uniprotMap = UniprotIdMap(source, target, retry, delay, contact,
returnNone)
MultiCachedDict.__init__(self, None, [dbMap, uniprotMap])
def __init__(self, dbPath, email):
ncbi = ProtId2ProtNameMap(email, useCache=False)
database = SqliteCache(filePath=dbPath, indict=None, table="gi2name",
key="gi", value="name")
MultiCachedDict.__init__(self, None, [database, ncbi])