def doGeneLookup(self, geneName, taxId, reviewed=False):
""" """
rL = []
try:
baseUrl = self.__urlPrimary
endPoint = "uniprot"
# hL = [("Accept", "application/xml")]
hL = []
if reviewed:
pD = {
"query":
'gene:"%s" and taxonomy:%s and reviewed:yes' %
(geneName, taxId),
"format":
"list"
}
else:
pD = {
"query": 'gene:"%s" and taxonomy:%s' % (geneName, taxId),
"format": "list"
}
ureq = UrlRequestUtil()
rspTxt, retCode = ureq.get(baseUrl, endPoint, pD, headers=hL)
tValL = rspTxt.split("\n") if rspTxt else []
idList = [tVal for tVal in tValL if tVal]
return idList, retCode
except Exception as e:
logger.exception("Failing with %s", str(e))
return rL, None
def testUnpBatchFetchGetUrllib(self):
"""UniProt batch fetch (uploadlists) get test (urllib)"""
baseUrl = "https://www.uniprot.org"
# baseUrl = "https://pir3.uniprot.org"
endPoint = "uploadlists"
idList = self.__unpIdList1[:10]
try:
# hD = {"Accept": "application/xml"}
hL = [("Accept", "application/xml")]
pD = {
"from": "ACC+ID",
"to": "ACC",
"format": "xml",
"query": " ".join(idList)
}
ureq = UrlRequestUtil()
# using wrapped version
ret, retCode = ureq.get(baseUrl,
endPoint,
pD,
headers=hL,
sslCert="enable")
logger.debug("XML result %r", ret)
nm = ret.count("<entry ")
logger.info("Result count %d status code %r", nm, retCode)
self.assertGreaterEqual(nm, len(idList))
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def __doRequestPrimary(self, idList):
"""
http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&retmode=xml&id=ID1,ID2,...
"""
baseUrl = "http://eutils.ncbi.nlm.nih.gov"
endPoint = "entrez/eutils/efetch.fcgi"
hL = [("Accept", "application/xml")]
pD = {"db": "pubmed", "retmode": "xml", "id": ",".join(idList)}
ureq = UrlRequestUtil()
return ureq.get(baseUrl, endPoint, pD, headers=hL)
def __doRequestSecondary(self, idList):
baseUrl = self.__urlSecondary
endPoint = "proteins/api/proteins"
#
hL = [("Accept", "application/xml")]
pD = {}
pD["size"] = "-1"
pD["accession"] = ",".join(idList)
ureq = UrlRequestUtil()
return ureq.get(baseUrl, endPoint, pD, headers=hL)
def __doRequestPrimary(self, idList):
""" """
baseUrl = self.__urlPrimary
endPoint = "uploadlists"
hL = [("Accept", "application/xml")]
pD = {
"from": "ACC+ID",
"to": "ACC",
"format": "xml",
"query": " ".join(idList)
}
ureq = UrlRequestUtil()
return ureq.get(baseUrl, endPoint, pD, headers=hL)
def getStatusDetails(self):
try:
version = releaseDateString = None
baseUrl = "https://www.ebi.ac.uk"
endPoint = "chembl/api/data/status.json"
hL = []
pD = {}
ureq = UrlRequestUtil()
ret, retCode = ureq.get(baseUrl, endPoint, pD, headers=hL, returnContentType="JSON")
logger.info("retCode %r ret %r", retCode, ret)
if ret:
tS = ret["chembl_db_version"] if ret and "chembl_db_version" in ret else None
version = tS.split("_")[1] if tS and tS.split("_")[1] else None
releaseDateString = ret["chembl_release_date"] if "chembl_release_date" in ret else None
except Exception as e:
logger.exception("Failing with %s", str(e))
return version, releaseDateString
def testUnpBatchFetchFail(self):
"""UniProt batch fetch (proteins) get test (expected failure)"""
baseUrl = "https://www0.ebi.ac.uk"
endPoint = "proteins/api/proteins"
idList = self.__unpIdList1[:10]
try:
hL = [("Accept", "application/xml")]
pD = {}
pD["size"] = "-1"
pD["accession"] = ",".join(idList)
ureq = UrlRequestUtil()
ret, retCode = ureq.get(baseUrl, endPoint, pD, headers=hL)
logger.debug("XML result %r", ret)
logger.debug("Result status code %r", retCode)
self.assertEqual(ret, None)
self.assertEqual(retCode, None)
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def testUnpBatchFetchGetEbi(self):
"""UniProt batch fetch (proteins) get test (EBI endpoint)"""
baseUrl = "https://www.ebi.ac.uk"
endPoint = "proteins/api/proteins"
idList = self.__unpIdList1[:10]
try:
hL = [("Accept", "application/xml")]
pD = {}
pD["size"] = "-1"
pD["accession"] = ",".join(idList)
ureq = UrlRequestUtil()
ret, retCode = ureq.get(baseUrl, endPoint, pD, headers=hL)
logger.debug("XML result %r", ret)
nm = ret.count("<entry ")
logger.info("Result count %d status code %r", nm, retCode)
self.assertGreaterEqual(nm, len(idList) - 1)
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def testNcbiFetchEntryPost(self):
"""NCBI batch fetch (efetch) get test"""
idList = ["AP012306.1", "U53879.1"]
database = "Nucleotide"
baseUrl = "https://eutils.ncbi.nlm.nih.gov"
endPoint = "entrez/eutils/efetch.fcgi"
try:
hL = [("Accept", "application/xml")]
pD = {}
pD["db"] = database
pD["id"] = ",".join(idList)
pD["retmode"] = "xml"
ureq = UrlRequestUtil()
ret, retCode = ureq.get(baseUrl, endPoint, pD, headers=hL)
nm = ret.count("<GBSeq_length>")
logger.debug("XML result %r", ret)
logger.info("Result count %d status code %r", nm, retCode)
self.assertGreaterEqual(nm, len(idList))
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def doLookup(self, itemList, itemKey="GENENAME"):
""" """
rL = []
try:
baseUrl = self.__urlPrimary
endPoint = "uploadlists"
# hL = [("Accept", "application/xml")]
hL = []
pD = {
"from": itemKey,
"to": "ACC",
"format": "list",
"query": " ".join(itemList)
}
ureq = UrlRequestUtil()
rspTxt, retCode = ureq.get(baseUrl, endPoint, pD, headers=hL)
tValL = rspTxt.split("\n") if rspTxt else []
idList = [tVal for tVal in tValL if tVal]
return idList, retCode
except Exception as e:
logger.exception("Failing with %s", str(e))
return rL, None
def testPubChemFetchClassification(self):
"""PubChem fetch classification test - can timeout"""
idTupList = [("2244", 200, "2244", "record"),
("123631", 200, "123631", "record"),
("2244", 200, "2244", "classification"),
("123631", 200, "123631", "classification")]
nameSpace = "cid"
domain = "compound"
searchType = "lookup"
# returnType = "record"
requestType = "GET"
outputType = "JSON"
baseUrl = "https://pubchem.ncbi.nlm.nih.gov"
httpCodesCatch = [404]
try:
for (identifier, testRetCode, testPcId, returnType) in idTupList:
for requestType in ["GET", "POST"]:
logger.info(
"namespace %r identifier %r returnType %r requestType %r",
nameSpace, identifier, returnType, requestType)
ret, retCode = None, None
pD = {}
hL = []
ureq = UrlRequestUtil()
if nameSpace in [
"cid", "name", "inchikey"
] and returnType in ["record"] and searchType in [
"lookup"
] and requestType == "GET":
uId = quote(identifier.encode("utf8"))
endPoint = "/".join([
"rest", "pug", domain, nameSpace, uId, outputType
])
ret, retCode = ureq.get(baseUrl,
endPoint,
pD,
headers=hL,
httpCodesCatch=httpCodesCatch,
returnContentType="JSON")
elif nameSpace in [
"cid", "name", "inchikey"
] and returnType in ["record"] and searchType in [
"lookup"
] and requestType == "POST":
endPoint = "/".join(
["rest", "pug", domain, nameSpace, outputType])
pD = {nameSpace: identifier}
ret, retCode = ureq.post(baseUrl,
endPoint,
pD,
headers=hL,
httpCodesCatch=httpCodesCatch,
returnContentType="JSON")
#
elif nameSpace in ["cid"] and returnType in [
"classification"
] and searchType in ["lookup"] and requestType == "GET":
# Needs to be specifically targeted on a particular compound ...
uId = quote(identifier.encode("utf8"))
endPoint = "/".join([
"rest", "pug", domain, nameSpace, uId, returnType,
outputType
])
# pD = {"classification_type": "simple"}
pD = {}
# pD = {nameSpace: identifier}
ret, retCode = ureq.getUnWrapped(
baseUrl,
endPoint,
pD,
headers={},
httpCodesCatch=httpCodesCatch,
returnContentType="JSON")
#
elif nameSpace in ["cid"] and returnType in [
"classification"
] and searchType in ["lookup"] and requestType == "POST":
# Needs to be specifically targeted on a particular compound ...
endPoint = "/".join([
"rest", "pug", domain, nameSpace, returnType,
outputType
])
# This is a long request return server codes may be observed 500
# pD = {nameSpace: identifier, "classification_type": "simple"}
pD = {nameSpace: identifier}
ret, retCode = ureq.postUnWrapped(
baseUrl,
endPoint,
pD,
headers={},
httpCodesCatch=httpCodesCatch,
returnContentType="JSON")
#
#
logger.debug("Result status code %r", retCode)
self.assertEqual(retCode, testRetCode)
if retCode == 200 and returnType == "record":
pcId = str(ret["PC_Compounds"][0]["id"]["id"]["cid"])
self.assertEqual(pcId, testPcId)
#
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()