def findDeltaPairs(dtaList, delta, ppm=5, intEp=20):
precMassArr = np.zeros((len(dtaList), 2))
pairs = []
for i in range(len(dtaList)):
precMassArr[i] = [DataFile.getPrecMassAndCharge(dtaList[i])[0], DataFile.getScanNum(dtaList[i])]
maxPrecMass = np.max(precMassArr, 0)[0]
epsilon = ppm * 10**-6 * maxPrecMass
resolution = epsilon/intEp
hPrecMassArr = copy.copy(precMassArr)
hPrecMassArr[:,0] = np.round(hPrecMassArr[:,0]/resolution)
hashedDict = {}
for elem in hPrecMassArr:
hInd = int(elem[0])
for hMass in range(hInd-intEp, hInd+intEp+1):
try:
hashedDict[hMass] += [(hMass-hInd, elem[1])]
except KeyError:
hashedDict[hMass] = [(hMass-hInd, elem[1])]
shiftHashDict = copy.copy(precMassArr)
shiftHashDict[:,0] = np.round((shiftHashDict[:,0] - delta)/resolution)
for i, elem in enumerate(shiftHashDict):
hInd = int(elem[0])
if hInd in hashedDict:
for possiblePair in hashedDict[hInd]:
if abs(possiblePair[0]) * resolution * 10**6/precMassArr[i][0] < ppm:
pairs += [(int(possiblePair[1]), int(elem[1]))]
return pairs
def findSamePrecMassClusters(dtaList, ppm=5):
precMassArr = np.zeros((len(dtaList), 2))
for i in range(len(dtaList)):
precMassArr[i] = [DataFile.getPrecMassAndCharge(dtaList[i])[0], DataFile.getScanNum(dtaList[i])]
precMassArr = precMassArr[np.argsort(precMassArr[:,0])]
clusters = [[i] for i in range(precMassArr.shape[0])]
i = 0
while i < len(clusters):
mergeClusters = False
epsilon = ppm * 10**-6 * precMassArr[clusters[i][0]][0]
for precMassInd1 in clusters[i]:
for precMassInd2 in clusters[i - 1]:
if (np.abs(precMassArr[precMassInd1][0] - precMassArr[precMassInd2][0]) < epsilon):
mergeClusters = True
break
if mergeClusters:
clusters[i - 1].extend(clusters[i])
del clusters[i]
else:
i = i + 1
scanFClusters = []
for cluster in clusters:
scanFClusters += [[precMassArr[i][1] for i in cluster]]
return scanFClusters
def getScanFDict(dtaList):
scanFDict = {}
for dta in dtaList:
scanF = DataFile.getScanNum(dta)
precMass = DataFile.getPrecMassAndCharge(dta)[0]
scanFDict[scanF] = {'dta': dta, 'precMass': precMass, 'sequenced': False}
return scanFDict
def getScanFDict(dtaList):
scanFDict = {}
for dta in dtaList:
scanF = DataFile.getScanNum(dta)
precMass = DataFile.getPrecMassAndCharge(dta)[0]
scanFDict[scanF] = {"dta": dta, "precMass": precMass, "sequenced": False}
return scanFDict