def testPubChemFetch(self):
"""PubChem fetch test"""
idTupList = [("JTOKYIBTLUQVQV-FGHQGBLESA-N", 404, None),
("CXHHBNMLPJOKQD-UHFFFAOYSA-N", 200, 78579)]
nameSpace = "inchikey"
domain = "compound"
searchType = "lookup"
returnType = "record"
requestType = "GET"
outputType = "JSON"
baseUrl = "https://pubchem.ncbi.nlm.nih.gov"
httpCodesCatch = [404]
try:
for (identifier, testRetCode, testPcId) in idTupList:
for requestType in ["GET", "POST"]:
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.getUnWrapped(
baseUrl,
endPoint,
pD,
headers=hL,
httpCodesCatch=httpCodesCatch,
returnContentType="JSON",
sslCert="enable")
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.postUnWrapped(
baseUrl,
endPoint,
pD,
headers=hL,
httpCodesCatch=httpCodesCatch,
returnContentType="JSON",
sslCert="enable")
#
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 = {nameSpace: identifier}
ret, retCode = ureq.getUnWrapped(
baseUrl,
endPoint,
pD,
headers=hL,
httpCodesCatch=httpCodesCatch,
returnContentType="JSON",
sslCert="enable")
#
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=hL,
httpCodesCatch=httpCodesCatch,
returnContentType="JSON",
sslCert="enable")
#
#
logger.debug("Result status code %r", retCode)
self.assertEqual(retCode, testRetCode)
if retCode == 200:
pcId = ret["PC_Compounds"][0]["id"]["id"]["cid"]
self.assertEqual(pcId, testPcId)
#
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
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()
def __fetchDescriptors(self, ccIdList, ccidxP, chunkSize=100):
"""Fetch transformed SMILES descriptors from the ChemAxon webservice.
Args:
ccIdList (list, str): chemical component identifier list
ccidxP (object): instance of the ChemCompIndexProvider()
chunksize (int, optional): number of SMILES per request. Defaults to 100.
Returns:
(dict): dictionary {<ccId>: [<transformed SMILES>, ...], ...}
Example API parameter data:
{
"errorHandlingMode": "FAIL_ON_ERROR",
"inputParams": "smiles",
"outputParams": "smiles",
"structures": [
"CC(C)[C@H](N)C=O",
"CC[C@H](C)[C@H](N)C=O",
"CC(C)C[C@H](N)C=O"
]
}
Example query:
curl -X POST "https://jchem-microservices.chemaxon.com/jwsio/rest-v1/molconvert/batch" -H "accept: */*"
-H "Content-Type: application/json" -d "{ \"errorHandlingMode\": \"FAIL_ON_ERROR\", \"inputParams\": \"smiles\",
\"outputParams\": \"mrv\", \"structures\": [ \"CC(C)[C@H](N)C=O\", \"CC[C@H](C)[C@H](N)C=O\", \"CC(C)C[C@H](N)C=O\" ]}"
"""
descrD = {}
smilesCcIdD = {}
smilesD = {}
for ccId in ccIdList:
smiL = list(
set(
ccidxP.getSMILES(ccId,
smiTypeList=[
"oe-iso-smiles", "oe-smiles",
"cactvs-iso-smiles", "cactvs-smiles"
])))
smilesCcIdD.setdefault(ccId, []).extend(smiL)
for smi in smiL:
smilesD.setdefault(smi, []).append(ccId)
#
logger.info("Translating (%d) SMILES for components (%d)",
len(smilesD), len(smilesCcIdD))
# ----
smiLL = [
list(smilesD.keys())[i:i + chunkSize]
for i in range(0, len(smilesD), chunkSize)
]
# ---
baseUrl = "https://jchem-microservices.chemaxon.com"
endPoint = "jwsio/rest-v1/molconvert/batch"
# hL = [("Accept", "application/json"), ("Content-Type", "application/json")]
hD = {"Accept": "application/json", "Content-Type": "application/json"}
try:
pD = {
"errorHandlingMode": "SKIP_ERROR",
"inputParams": "smiles",
"outputParams": "smiles"
}
#
iCount = 0
for smiL in smiLL:
iCount += 1
ureq = UrlRequestUtil()
pD["structures"] = smiL
logger.debug("pD %r", pD)
rDL, retCode = ureq.postUnWrapped(
baseUrl,
endPoint,
pD,
headers=hD,
sendContentType="application/json",
returnContentType="application/json")
logger.debug("API result (%r) %r", retCode, rDL)
if rDL and len(rDL) == len(smiL):
for ii, rD in enumerate(rDL):
if "structure" in rD and "successful" in rD and rD[
"successful"]:
if smiL[ii] == rD["structure"]:
continue
for ccId in smilesD[smiL[ii]]:
if ccId in descrD and rD[
"structure"] in descrD[ccId]:
continue
if rD["structure"] in smilesCcIdD[ccId]:
continue
descrD.setdefault(ccId,
[]).append(rD["structure"])
else:
logger.info("Chunk %d failed (%d)", iCount, len(rDL))
if iCount % 10 == 0:
logger.info("Completed processing chunk (%d/%d)", iCount,
len(smiLL))
#
except Exception as e:
logger.exception("Failing with %s", str(e))
return descrD
