def computeIndexFeatures(aaAlphabet, peptide, index, polarAa, hydrophobicAa):
features = []
aas = dm.getAminoAcidList(peptide)
features.append(indexSum(aas, index))
features.append(indexAvg(aas, index))
features.append(indexN(aas, index))
features.append(indexC(aas, index))
features.append(indexNearestNeighbour(aas, index, polarAa))
maxPartSum5, minPartSum5 = indexMaxMinPartialSum(aas, index, 5)
maxPartSum2, minPartSum2 = indexMaxMinPartialSum(aas, index, 2)
features.append(maxPartSum5)
features.append(maxPartSum2)
features.append(minPartSum5)
features.append(minPartSum2)
maxHsideHelix, minHsideHelix = indexMaxMinHydrophobicSideHelix(aas, index)
features.append(maxHsideHelix)
features.append(minHsideHelix)
maxHmoment100, minHmoment100 = indexMaxMinHydrophobicMoment(aas, index, 100, 11)
maxHmoment180, minHmoment180 = indexMaxMinHydrophobicMoment(aas, index, 180, 11)
features.append(maxHmoment100)
features.append(maxHmoment180)
features.append(minHmoment100)
features.append(minHmoment180)
features.append(indexSumSquaredDiff(aas, index))
features.append(numberTypeAA(aas, polarAa))
features.append(numberConsecTypeAA(aas, polarAa))
features.append(numberTypeAA(aas, hydrophobicAa))
features.append(numberConsecTypeAA(aas, hydrophobicAa))
return features
def computeRetentionFeatureMatrix(aaAlphabet, psmDescriptions, customIndex):
ptmsPresent = False
if hasPtms(aaAlphabet):
numFeatures = 20 + 1 + len(aaAlphabet)
ptmsPresent = True
else:
numFeatures = 20 + 20 + 2 + len(aaAlphabet)
featureMatrix = np.zeros((len(psmDescriptions),0))
if not ptmsPresent:
polarAa, hydrophobicAa = getExtremeRetentionAA(kyteDoolittleIndex)
kyteDoolittleFeatureMatrix = np.zeros((len(psmDescriptions), 20))
for i, psmd in enumerate(psmDescriptions):
kyteDoolittleFeatureMatrix[i] = computeIndexFeatures(aaAlphabet, psmd.peptide, kyteDoolittleIndex, polarAa, hydrophobicAa)
featureMatrix = np.concatenate((featureMatrix, kyteDoolittleFeatureMatrix), axis = 1)
polarAa, hydrophobicAa = getExtremeRetentionAA(customIndex)
customFeatureMatrix = np.zeros((len(psmDescriptions), 20))
for i, psmd in enumerate(psmDescriptions):
customFeatureMatrix[i] = computeIndexFeatures(aaAlphabet, psmd.peptide, customIndex, polarAa, hydrophobicAa)
featureMatrix = np.concatenate((featureMatrix, customFeatureMatrix), axis = 1)
if not ptmsPresent:
bulkinessFeatureVector = np.zeros((len(psmDescriptions), 1))
for i, psmd in enumerate(psmDescriptions):
aas = dm.getAminoAcidList(psmd.peptide)
bulkinessFeatureVector[i] = indexSum(aas, bulkinessIndex)
featureMatrix = np.concatenate((featureMatrix, bulkinessFeatureVector), axis = 1)
lengthFeatureVector = np.zeros((len(psmDescriptions), 1))
for i, psmd in enumerate(psmDescriptions):
aas = dm.getAminoAcidList(psmd.peptide)
lengthFeatureVector[i] = len(aas)
featureMatrix = np.concatenate((featureMatrix, lengthFeatureVector), axis = 1)
aaFeatureVector = computeRetentionIndexFeatureMatrix(aaAlphabet, psmDescriptions)
featureMatrix = np.concatenate((featureMatrix, aaFeatureVector), axis = 1)
normalizeFeatures(featureMatrix)
return featureMatrix
def getBOWfeature(file,num):
hydrophobicity =[]
peptide_length = []
psmDescriptions, aaAlphabet = el.processTrainData(file)
Index = rm.buildRetentionIndex(aaAlphabet,psmDescriptions,True)
customIndex = dict(zip(aaAlphabet, Index))
for i, psmd in enumerate(psmDescriptions):
aas = dm.getAminoAcidList(psmd.peptide)
peptide_length.append(len(psmd.peptide))
hydrophobicity.append(bowindexSum(aas, customIndex))
peptide,rt = dg.extract(psmDescriptions,len(psmDescriptions))
bow_feature = BOW(peptide,hydrophobicity,peptide_length,aaAlphabet,num)
return psmDescriptions ,bow_feature
def computeRetentionFeatureVector(aaAlphabet, peptide, customIndex):
ptmsPresent = False
if hasPtms(aaAlphabet):
numFeatures = 20 + 1 + len(aaAlphabet)
ptmsPresent = True
else:
numFeatures = 20 + 20 + 2 + len(aaAlphabet)
featureVector = np.zeros((1, 0))
if not ptmsPresent:
polarAa, hydrophobicAa = getExtremeRetentionAA(kyteDoolittleIndex)
kyteDoolittleFeatureVector = np.zeros((1, 20))
kyteDoolittleFeatureVector[:] = computeIndexFeatures(
aaAlphabet, peptide, kyteDoolittleIndex, polarAa, hydrophobicAa
)
featureVector = np.concatenate((featureVector, kyteDoolittleFeatureVector), axis=1)
polarAa, hydrophobicAa = getExtremeRetentionAA(customIndex)
customFeatureVector = np.zeros((1, 20))
customFeatureVector[:] = computeIndexFeatures(aaAlphabet, peptide, customIndex, polarAa, hydrophobicAa)
featureVector = np.concatenate((featureVector, customFeatureVector), axis=1)
if not ptmsPresent:
bulkinessFeatureVector = np.zeros((1, 1))
aas = dm.getAminoAcidList(peptide)
bulkinessFeatureVector[:] = indexSum(aas, bulkinessIndex)
featureVector = np.concatenate((featureVector, bulkinessFeatureVector), axis=1)
lengthFeatureVector = np.zeros((1, 1))
aas = dm.getAminoAcidList(peptide)
lengthFeatureVector[:] = len(aas)
featureVector = np.concatenate((featureVector, lengthFeatureVector), axis=1)
aaFeatureVector = computeRetentionIndexFeatureMatrix(aaAlphabet, peptide)
featureVector = np.concatenate((featureVector, aaFeatureVector), axis=1)
normalizeFeatures(featureVector)
return np.array(featureVector[0])
def computeRetentionIndexFeatures(aaAlphabet, peptide):
aas = dm.getAminoAcidList(peptide)
featureVector = np.zeros((1, len(aaAlphabet)))
for aa in aas:
featureVector[0][aaAlphabet.index(aa)] += 1
return featureVector
