def calculateAvDlossPerDose(PDBmulti, n):
# calculated the average Dloss metric over all atoms of each residue/nucleotide type within
# TRAP complex
# determine full list of residue types present
uniq_resis = []
for atom in PDBmulti:
if atom.basetype not in uniq_resis:
uniq_resis.append(atom.basetype)
uniq_resis.sort()
x = np.array(uniq_resis)
# for each residue type, find average Dloss at each dose
# first group atoms by residue type:
atomsByRes = {}
for res in uniq_resis:
atomsByRes[res] = []
for atom in PDBmulti:
atomsByRes[atom.basetype].append(atom.mindensity[0:9])
# calculate average Dloss over structure for each dose
avDlossOverall = np.mean([atom.mindensity[0:9] for atom in PDBmulti], 0)
print "Average Dloss for structure per dose as following:"
print ",".join([str(val) for val in avDlossOverall])
# for each residue type, calculate mean Dloss at each dose
avDlossByRes = {}
stdDlossByRes = {}
nthTileDlossByRes = {}
confIntDlossByRes = {}
avDlossByResNorm = {}
for key in atomsByRes.keys():
avDloss = np.mean(atomsByRes[key], 0)
stdDloss = np.std(atomsByRes[key], 0)
nthTileDloss = np.percentile(atomsByRes[key], n, 0)
avDlossNorm = avDloss - avDlossOverall
# calculate 95% confidence interval
confIntDloss = []
for i in range(0, 9):
confIntDloss.append(mean_confidence_interval([val[i] for val in atomsByRes[key]]))
avDlossByRes[key] = avDloss
stdDlossByRes[key] = stdDloss
nthTileDlossByRes[key] = nthTileDloss
confIntDlossByRes[key] = confIntDloss
avDlossByResNorm[key] = avDlossNorm
# get results
csvOutput_stdError = open("avDlossPerDataset_stdError.csv", "w")
csvOutput_ConfIntError = open("avDlossPerDataset_confIntError.csv", "w")
csvOutput_nthTile = open("{}thTile_DlossPerDataset.csv".format(n), "w")
csvOutput_DlossNormalised = open("avDlossPerDataset_DlossNormalised.csv", "w")
for key in avDlossByRes.keys():
avDlossList = list([str(element) for element in avDlossByRes[key]])
stdDlossList = list([str(element) for element in stdDlossByRes[key]])
nthTileDlossList = list([str(element) for element in nthTileDlossByRes[key]])
confIntDlossList = list([str(element) for element in confIntDlossByRes[key]])
avDlossByResNormList = list([str(element) for element in avDlossByResNorm[key]])
print "\n***\n{}: {}".format(key, "-->".join(avDlossList))
print "{}: {}".format(key, "-->".join(stdDlossList))
print "{}: {}".format(key, "-->".join(confIntDlossList))
csvOutput_stdError.write("{},{}\n".format(key, ",".join(avDlossList)))
csvOutput_stdError.write("{},{}\n".format(key, ",".join(stdDlossList)))
csvOutput_nthTile.write("{},{}\n".format(key, ",".join(nthTileDlossList)))
csvOutput_DlossNormalised.write("{},{}\n".format(key, ",".join(avDlossByResNormList)))
csvOutput_ConfIntError.write("{},{}\n".format(key, ",".join(avDlossList)))
csvOutput_ConfIntError.write("{},{}\n".format(key, ",".join(confIntDlossList)))
csvOutput_stdError.close()
csvOutput_ConfIntError.close()
csvOutput_nthTile.close()
csvOutput_DlossNormalised.close()
def densMetricErrorbarGraphs(self,auto,where,metricTypes,confInt):
# function to plot density change as function of dataset number
# for a specific atom in the structure, with the mean value over
# all protein chains plotted, along with error bars for the 22
# equivalent atoms present. specify auto=False to specify atom type
# on the command line
# 'where' specifies where to plot, if doesn't exist, makes directory in
# current directory
# metricTypes takes values 1 or 2 if auto == True
# If 'confInt' is True then error bars are 95% confidence intervals,
# otherwise, 1 SD used at each dose
# get equivalent atoms of specified type (command line input to specify)
self.getEquivalentAtoms(auto)
# determine whether dealing with protein or RNA atoms
if self.equivAtoms[0].boundOrUnbound() in ('unbound protein','bound protein'):
protein = True
else:
protein = False
sns.set(style="white", context="talk")
f = plt.figure(figsize=(16, 8))
# define x range here (damage set numbers or doses if specified)
if self.doseList == []:
x = range(2,len(self.equivAtoms[0].meandensity)+2)[0:10]
x_label = 'Damage set'
else:
x = self.doseList
x_label = "Dose (MGy)"
# Determine density metric set to plot here
# Currently two distinct options given below
densMets1 = ['loss','net','mean','gain']
densMets2 = ['loss','net','mean','gain','bfactor','bdamage']
densMets3 = ['|loss|','loss','net','mean','gain','bfactor']
densMets4 = ['max-simple','median-simple','gain','median','mean','loss']
if auto == False:
print 'Which metrics would you like to plot...'
userInput = raw_input("1 or 2?: ")
if userInput == str(1):
densMets = densMets1
normTypes = ['Standard','Calpha normalised']
else:
densMets = densMets2
normTypes = ['Standard']
if auto == True:
if metricTypes == 1:
densMets = densMets1
normTypes = ['Standard','Calpha normalised']
elif metricTypes == 2:
densMets = densMets2
normTypes = ['Standard']
elif metricTypes == 3:
densMets = densMets3
normTypes = ['Standard']
elif metricTypes == 4:
densMets = densMets4
normTypes = ['Standard']
i = 0
HotellingTsquareDict = {}
for densMet in densMets:
for normType in normTypes:
if densMet in ('mean','gain') and normType in ('Calpha normalised'):
continue
i+=1
yValue = {}
# for protein atoms, group by bound and unbound chains
if protein == True:
for boundType in ('unbound','bound'):
yValue[boundType] = {}
for valType in ('mean','std','95ConfInt'):
yValue[boundType][valType] = []
for j in range(0,len(x)):
yValue[boundType]['mean'].append(np.mean([atom.densMetric[densMet][normType]['values'][j] for atom in self.equivAtoms if atom.boundOrUnbound() == '{} protein'.format(boundType)]))
yValue[boundType]['std'].append(np.std([atom.densMetric[densMet][normType]['values'][j] for atom in self.equivAtoms if atom.boundOrUnbound() == '{} protein'.format(boundType)]))
yValue[boundType]['95ConfInt'].append(mean_confidence_interval([atom.densMetric[densMet][normType]['values'][j] for atom in self.equivAtoms if atom.boundOrUnbound() == '{} protein'.format(boundType)]))
# for RNA atoms, just create a list of density values
if protein == False:
yValue['RNA 1'] = {}
yValue['RNA 2'] = {}
for valType in ('mean','std'):
yValue['RNA 1'][valType] = []
yValue['RNA 2'][valType] = []
for j in range(0,len(x)):
yValue['RNA 1']['mean'].append(np.mean([atom.densMetric[densMet][normType]['values'][j] for atom in self.equivAtoms if atom.residuenum%5 == self.residueNum]))
yValue['RNA 1']['std'].append(np.std([atom.densMetric[densMet][normType]['values'][j] for atom in self.equivAtoms if atom.residuenum%5 == self.residueNum]))
yValue['RNA 2']['mean'].append(np.mean([atom.densMetric[densMet][normType]['values'][j] for atom in self.equivAtoms if atom.residuenum%5 != self.residueNum]))
yValue['RNA 2']['std'].append(np.std([atom.densMetric[densMet][normType]['values'][j] for atom in self.equivAtoms if atom.residuenum%5 != self.residueNum]))
ax = plt.subplot(2,3,i)
ax.set_xlim([0, 29])
if protein == True:
if confInt == True:
plt.errorbar(x,yValue['unbound']['mean'],yerr=yValue['unbound']['95ConfInt'], fmt='-o',capthick=2,color='#99ccff',label='Non-bound')
plt.errorbar(x,yValue['bound']['mean'],yerr=yValue['bound']['95ConfInt'],fmt='-o',capthick=2,color='#f47835',label='Bound')
else:
plt.errorbar(x,yValue['unbound']['mean'],yerr=yValue['unbound']['std'], fmt='-o',capthick=2,color='#99ccff',label='Non-bound')
plt.errorbar(x,yValue['bound']['mean'],yerr=yValue['bound']['std'],fmt='-o',capthick=2,color='#f47835',label='Bound')
else:
try:
plt.errorbar(x,yValue['RNA 1']['mean'],yerr=yValue['RNA 1']['std'], fmt='-o',capthick=2,color='r',label='G1')
plt.errorbar(x,yValue['RNA 2']['mean'],yerr=yValue['RNA 2']['std'], fmt='-o',capthick=2,color='g',label='G3')
except KeyError:
plt.errorbar(x,yValue['RNA 1']['mean'],yerr=yValue['RNA 1']['std'], fmt='-o',capthick=2,color='r')
# ax.set_xlim([1, 11])
ax.legend(loc='best')
plt.xlabel(x_label)
if normType == 'Calpha normalised':
plt.ylabel('Normalised D{} change'.format(densMet))
else:
plt.ylabel('{} D{} change'.format(normType,densMet))
# # perform Hotelling T-squared test if protein atoms
# if protein == True:
# keyVal = '{} D{}'.format(normType,densMet)
# HotellingTsquareDict[keyVal] = {}
# F,p_value,reject = self.hotellingTsquareTest(densMet,normType) # run Hotelling's T squared test to distinguish between bound and unbound TRAP rings
# HotellingTsquareDict[keyVal]['F value'] = F
# HotellingTsquareDict[keyVal]['p value'] = p_value
# HotellingTsquareDict[keyVal]['reject?'] = reject
plt.subplots_adjust(top=0.90)
f.subplots_adjust(hspace=0.4)
f.subplots_adjust(wspace=0.5)
f.suptitle('damage metrics vs damage set: {} {} {}'.format(atom.basetype,atom.residuenum,atom.atomtype),fontsize=20)
# check if directory exists to save graphs to and make if not:
if not os.path.exists(where):
os.makedirs(where)
# save graphs to directly specified by 'where'
if confInt == True:
f.savefig('{}/6DamageSubplots_{}_{}_{}_95confIntErrorbars.png'.format(where,atom.basetype,atom.residuenum,atom.atomtype))
else:
f.savefig('{}/6DamageSubplots_{}_{}_{}_SDerrorbars.png'.format(where,atom.basetype,atom.residuenum,atom.atomtype))