def multiCore_PPILinkPred_shared(samplePPIbr,
scoringMethod,
scoreArgsDict,
coreNo,
topNo=None,
logging=False,
nodePairs=None):
# @param scoreArgs: dict, assign the normalization functions (normFunc, uvSpec, xySpec, uvContrib, xyContrib, dualCN)
normOrder = [
'normFunc', 'uvSpec', 'xySpec', 'uvContrib', 'xyContrib', 'dualCN',
'uvJoin'
]
scoreArgs = [
'null' if normTag not in scoreArgsDict else scoreArgsDict[normTag]
for normTag in normOrder
]
samplePPIr = ns.BRToRelat(ns.toDualBR(samplePPIbr), rSet=True)
sampleNodes = ns.BRToNode(samplePPIbr)
if nodePairs is None: nodePairs = list(combinations(sampleNodes, 2))
splitStartIndex = [
i * math.floor(len(nodePairs) / coreNo) for i in range(0, coreNo)
] # both splitting is correct
splitEndIndex = [(i + 1) * math.floor(len(nodePairs) / coreNo)
if i != coreNo - 1 else len(nodePairs)
for i in range(0, coreNo)]
mgr, dataMgr = Manager(), Manager()
PPIdataQ, PPIresQ = dataMgr.Queue(), mgr.Queue()
if logging: logging = [True if i == 0 else False for i in range(coreNo)]
else: logging = [False for i in range(coreNo)]
for i in range(len(splitStartIndex)):
PPIdataQ.put(nodePairs[splitStartIndex[i]:splitEndIndex[i]])
nodePairs = None
args = (PPIdataQ, samplePPIr, scoringMethod, scoreArgs, logging, PPIresQ)
func = partial(_multiCore_handler_shared, args)
with Pool(coreNo) as p:
p.map(func, [i for i in range(coreNo)])
if logging: print("\n")
mergedScores, mergedPPIbrs = [], []
PPIresL = [PPIresQ.get() for i in range(coreNo)]
for [predictedPPIbr, scores] in PPIresL:
mergedScores += scores
mergedPPIbrs += predictedPPIbr
sortedPPIbrs, sortedScores = hr.sort_key_val(mergedPPIbrs, mergedScores)
if topNo is None: topNo = len(sortedPPIbrs)
topPredPPIbrs = sortedPPIbrs[0:topNo]
topScores = sortedScores[0:topNo]
return topPredPPIbrs, topScores
def PPILinkPred(samplePPIbr):
samplePPIr = ns.BRToRelat(ns.toDualBR(samplePPIbr), rSet=True)
sampleNodes = ns.BRToNode(samplePPIbr)
nodePairs = list(combinations(sampleNodes, 2))
scores, predictedPPIbrs = [], []
for nodePair in nodePairs:
[nodeX, nodeY] = nodePair
if nodeY in samplePPIr[nodeX]: continue
score = L3_normalization(samplePPIr, nodeX, nodeY)
scores.append(score)
predictedPPIbrs.append(nodePair)
sortedPPIbrs, sortedScores = hr.sort_key_val(predictedPPIbrs, scores)
return sortedPPIbrs, sortedScores
def STRING_homo_HPC_combine():
dataset_len = int(
len([
*string.parse_STRING(
ppiFile='./data/STRING/9606.protein.links.v11.0.txt',
typeFile='./data/STRING/9606.protein.actions.v11.0.txt',
uniProtMap=
'./data/UniProt/uniprot-taxonomy_9606_STRING.tab',
root='../',
wFile_GGI='./data/parsed/STRING_homo_GGI.pkl',
wFile_PPI='./data/parsed/STRING_homo_PPI.pkl')
][1].index) * 0.5)
dataset = "STRING"
datasetName = "STRING_homo"
for folder in os.listdir("./resultData/h**o/"):
if dataset in folder:
for trial in range(0, 10):
print("folder: {}, trial: {}".format(folder, trial))
filenames = os.listdir("./resultData/h**o/" + folder)
ppiFiles = [i for i in filenames if "PPI" in i]
curTrialFiles = [
i for i in ppiFiles if i.split("_")[-3] == str(trial)
]
topPPIs, topScores = [], []
for i in curTrialFiles:
with open("./resultData/h**o/{}/{}".format(folder, i),
"r") as f:
topPPIs += json.loads(f.read())
with open(
"./resultData/h**o/{}/{}".format(
folder, "_".join(i.split("_")[:-1]) +
"_score.json"), "r") as f:
topScores += json.loads(f.read())
# sort
topPPIs, topScores = hr.sort_key_val(
topPPIs, topScores)
topPPIs = topPPIs[0:dataset_len]
topScores = topScores[0:dataset_len]
with open(
"./resultData/h**o/{}/{}_{}_trimmedPPIs.json".
format(folder, curTrialFiles[0].split("_")[0],
datasetName), "a+") as f:
f.write(json.dumps(topPPIs) + "\n")
with open(
"./resultData/h**o/{}/{}_{}_trimmedScores.json".
format(folder, curTrialFiles[0].split("_")[0],
datasetName), "a+") as f:
f.write(json.dumps(topScores) + "\n")
def bioGRID_homo_HPC_combine():
dataset_len = int(
len([
*bg.parse_bioGRID(
filename=
'./data/BioGRID/BIOGRID-ORGANISM-Homo_sapiens-3.5.187.tab2.txt',
wFile_GGI='./data/parsed/BioGRID_homo_GGI.pkl',
wFile_PPI='./data/parsed/BioGRID_homo_PPI.pkl',
root="../")
][1].index) * 0.5)
dataset = "bioGRID"
datasetName = "bioGRID_homo"
for folder in os.listdir("./resultData/h**o/"):
if dataset in folder:
for trial in range(6, 10):
print("folder: {}, trial: {}".format(folder, trial))
filenames = os.listdir("./resultData/h**o/" + folder)
ppiFiles = [i for i in filenames if "PPI" in i]
curTrialFiles = [
i for i in ppiFiles if i.split("_")[-3] == str(trial)
]
topPPIs, topScores = [], []
for i in curTrialFiles:
with open("./resultData/h**o/{}/{}".format(folder, i),
"r") as f:
topPPIs += json.loads(f.read())
with open(
"./resultData/h**o/{}/{}".format(
folder, "_".join(i.split("_")[:-1]) +
"_score.json"), "r") as f:
topScores += json.loads(f.read())
# sort
topPPIs, topScores = hr.sort_key_val(
topPPIs, topScores)
topPPIs = topPPIs[0:dataset_len]
topScores = topScores[0:dataset_len]
with open(
"./resultData/h**o/{}/{}_{}_trimmedPPIs.json".
format(folder, curTrialFiles[0].split("_")[0],
datasetName), "a+") as f:
f.write(json.dumps(topPPIs) + "\n")
with open(
"./resultData/h**o/{}/{}_{}_trimmedScores.json".
format(folder, curTrialFiles[0].split("_")[0],
datasetName), "a+") as f:
f.write(json.dumps(topScores) + "\n")