def getParetoInfo(clusterVector = None, extraData = None):
paretoInfo = []
extraDataWithClusters = extraData[:]
if clusterVector and extraData:
for i in range(len(clusterVector)):
clusterProvisionalList = []
dataList = []
clusterName = "Cluster #%i" % i
for clusterElement in clusterVector[i]:
nValue = float("nan")
extraDataIndex = stats.firstIndex(stats.extractColumns(extraDataWithClusters, 0), clusterElement)
if str(extraData[extraDataIndex][1]).lower() != "nan":
nValue = int(extraData[extraDataIndex][1])
clusterProvisionalList.append([extraDataIndex,
nValue, # n
float(extraData[extraDataIndex][4]), #X
extraData[extraDataIndex][0]]) # id
dataList.append([nValue,
abs(float(extraData[extraDataIndex][4]))])
for clusterElement in clusterProvisionalList:
dataPoint = [clusterElement[1], abs(clusterElement[2])] # {n, X}
extraDataWithClusters[clusterElement[0]].extend([clusterName, stats.isParetoFront(dataPoint, dataList)])
paretoInfo.append([extraDataWithClusters[clusterElement[0]][0], extraDataWithClusters[clusterElement[0]][6]])
return paretoInfo, extraDataWithClusters
def filterNFDRorZ(lowerData,
higherData,
minN=2,
maxN=1e6,
minZ=0.0,
maxFDR=sys.float_info.max,
useFDR=True):
# filter n
# listNok, because listNok is for faster searching,
# while listNok2 is for storing the nij
listNok = []
listNok2 = []
for row in lowerData:
lowerElement = convertToStatsResult(row)
if lowerElement.nij >= minN and lowerElement.nij <= maxN and lowerElement.id2 not in listNok:
listNok.append(lowerElement.id2)
listNok2.append([lowerElement.id2, lowerElement.nij])
# filter Z
listNZok = []
for row in higherData:
higherElement = convertToStatsResult(row)
if useFDR:
if str(higherElement.FDRij
) != "nan" and higherElement.id1 in listNok:
if higherElement.FDRij <= maxFDR:
index = stats.firstIndex(listNok, higherElement.id1)
nij = int(listNok2[index][1])
listNZok.append([
higherElement.id1, nij, higherElement.Zij,
higherElement.FDRij,
higherElement.Xi - higherElement.Xj
])
else:
if str(higherElement.Zij
) != "nan" and higherElement.id1 in listNok:
if abs(higherElement.Zij) >= minZ:
index = stats.firstIndex(listNok, higherElement.id1)
nij = int(listNok2[index][1])
listNZok.append([
higherElement.id1, nij, higherElement.Zij,
higherElement.FDRij,
higherElement.Xi - higherElement.Xj
])
return listNZok
def scalarProduct(vectorA, vectorB): # returns the percent of elements of A # that are contained in B # where 1 means all the elements of A are present in B # and 0 means A and B are orthogonal # next is just to speed up a bit the program if len(vectorA) < len(vectorB): vectorShort = vectorA vectorLong = vectorB else: vectorShort = vectorB vectorLong = vectorA product = 0 for element in vectorShort: first = stats.firstIndex(vectorLong, element) if first == -1: pass else: product += 1 return product
def associateElements(inStats = "", uFile = "", relFile = ""):
results = []
relations = stats.loadRelationsFile(relFile)
relations = stats.sortByIndex(relations, 0)
statsData = stats.loadStatsDataFile(inStats)
ZijList = []
for element in statsData:
ZijList.append([element[3], element[7]])
theorList = []
experList = []
N = len(ZijList)
for i in range(N):
theorList.append([ZijList[i][0], ZijList[i][1], norm.cdf(float(ZijList[i][1]))])
experList.append([ZijList[i][0], ZijList[i][1], (float(i) + 0.5) / float(N)])
higherElements = stats.load2stringList(uFile, removeCommas = True)
# WARNING! higherElements must be a list of lists
# with each sublist being id, n, Z, FDR, X
# begin: jmrc
if not higherElements:
sms = "ERROR: higherElements is empty. The higherElements must be a list of lists with each sublist being id, n, Z, FDR, X"
sys.exit(sms)
# end: jmrc
elementList = []
if higherElements[0] == ['id', 'Z', 'n']:
# this means the list comes from SanXoTSqueezer
# so the header and the extra columns have to be removed
for element in higherElements[1:]:
# switch to id, n, Z, FDR
elementList.append([element[0], element[2], element[1], float("nan"), float("nan")])
if higherElements[0] == ['id', 'n', 'Z', 'FDR']:
# this means it does not contain X, so a nan is put on its place
for element in higherElements[1:]:
elementList.append([element[0], element[1], element[2], element[3], float("nan")])
if higherElements[0] == ['id', 'n', 'Z', 'FDR', 'X']:
for element in higherElements[1:]:
elementList.append([element[0], element[1], element[2], element[3], element[4]])
# otherwise
if higherElements[0] != ['id', 'Z', 'n'] and higherElements[0] != ['id', 'n', 'Z', 'FDR'] and higherElements[0] != ['id', 'n', 'Z', 'FDR', 'X']:
for element in higherElements:
elementList.append([element[0], float("nan"), float("nan"), float("nan"), float("nan")])
statsData = stats.sortByIndex(statsData, 7)
relationsFirstColumn = stats.extractColumns(relations, 0)
relationsSecondColumn = stats.extractColumns(relations, 1)
experListFirstColumn = stats.extractColumns(experList, 0)
for uElement in elementList:
lowerElementList = []
first = stats.firstIndex(relationsFirstColumn, uElement[0])
if first > -1: # -1 means it is not in the list
notInList = 0
last = stats.lastIndex(relationsFirstColumn, uElement[0])
lowerElements = relationsSecondColumn[first:last + 1] # "+1" is to include the last one
for element in lowerElements:
lowerIndex = stats.firstIndex(experListFirstColumn, element)
if lowerIndex > -1: # -1 means it is not in the list
lowerElementList.append(element)
else:
notInList += 1
lowerElementList = stats.sortByIndex(lowerElementList, 0)
results.append([uElement[0], lowerElementList])
else:
if len(uElement[0].strip()) > 0:
results.append([uElement[0], None])
return results, elementList, ""