def bestInfo(self, id1, id2, alignment, coevolution):
"Points out the best coevolution scores"
seq = class_sequence(self.file1, self.file2, self.id1, self.id2,
self.chain1, self.chain2, self.parameterfile,
self.dirname)
histogram = LP(self.parameterfile, "results_histogram")
heatmap = LP(self.parameterfile, "results_heatmap")
best_info = LP(self.parameterfile, "best_results")
surface1 = []
surface2 = []
interface = []
try:
surface1 = seq.parseSurfacePDB(id1)
surface2 = seq.parseSurfacePDB(id2)
except:
pass
try:
interface = seq.parseInterfacePDB(id1)
except:
pass
input = self.dirname + alignment + "_" + coevolution + ".txt"
output = self.dirname + alignment + "_" + coevolution + "_best.txt"
bestResults(input, output, best_info, surface1, surface2, interface)
if histogram == True:
input = self.dirname + alignment + "_" + coevolution + ".txt"
output = self.dirname + alignment + "_" + coevolution + "_hg.png"
drawHistogram(input, output)
if heatmap == True:
input = self.dirname + alignment + "_" + coevolution + ".txt"
output = self.dirname + alignment + "_" + coevolution + "_hm.png"
drawHeatmap(id1, id2, input, output)
def bestInfo(self, id1, id2, alignment, coevolution):
"Points out the best coevolution scores"
seq = class_sequence(self.file1, self.file2, self.id1, self.id2,
self.chain1, self.chain2, self.parameterfile,
self.dirname)
histogram = LP(self.parameterfile, "results_histogram")
heatmap = LP(self.parameterfile, "results_heatmap")
best_info = LP(self.parameterfile, "best_results")
surface1 = []
surface2 = []
interface = []
try:
surface1 = seq.parseSurfacePDB(id1)
surface2 = seq.parseSurfacePDB(id2)
except:
pass
try:
interface = seq.parseInterfacePDB(id1)
except:
pass
input = self.dirname + alignment + "_" + coevolution + ".txt"
output = self.dirname + alignment + "_" + coevolution + "_best.txt"
bestResults(input, output, best_info, surface1, surface2, interface)
if histogram == True:
input = self.dirname + alignment + "_" + coevolution + ".txt"
output = self.dirname + alignment + "_" + coevolution + "_hg.png"
drawHistogram(input, output)
if heatmap == True:
input = self.dirname + alignment + "_" + coevolution + ".txt"
output = self.dirname + alignment + "_" + coevolution + "_hm.png"
drawHeatmap(id1, id2, input, output)
def coevolAnalysis(self, file1, file2, id1, id2, chain1, chain2, alignment,
coevolution):
"Returns a matrix of coevolution scores"
seq = class_sequence(self.file1, self.file2, self.id1, self.id2,
self.chain1, self.chain2, self.parameterfile,
self.dirname)
aln = class_alignment(self.id1, self.id2, self.alignment,
self.parameterfile, self.dirname)
alignment1 = aln.cutAlignment(file1, id1, alignment)
alignment2 = aln.cutAlignment(file2, id2, alignment)
try:
assert len(alignment1) == len(alignment2)
except:
raise StandardError, "Alignments must have the same number of sequences"
protein1 = []
protein2 = []
try:
protein1 = seq.matchResiduePosition(id1, chain1)
protein2 = seq.matchResiduePosition(id2, chain2)
except:
pass
info = dict()
alignment1 = [e for e in alignment1]
columns1 = transpose(alignment1)
alignment2 = [e for e in alignment2]
columns2 = transpose(alignment2)
if coevolution == "mi":
Flash('Mutual Information')
mi = dict()
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
mi[(i, j)] = mutualInformation(i, j, columns1, columns2,
pD1, pD2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(mi[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if mi[(i, j)] != 0.0:
info[(i, j)] = mi[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "mie":
Flash('Mutual Information by Pair Entropy')
mie = dict()
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
mie[(i, j)] = miEntropy(i, j, columns1, columns2, pD1, pD2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(mie[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if mie[(i, j)] != 0.0:
info[(i, j)] = mie[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "rcwmi":
Flash('Row and Column Weighed Mutual Information')
rcwmi = dict()
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
i_all = dict()
all_j = dict()
for i in range(len(columns1)):
v_i = 0
for j in range(len(columns2)):
v_i += mutualInformation(i, j, columns1, columns2, pD1,
pD2)
i_all[i] = v_i
for j in range(len(columns2)):
v_j = 0
for i in range(len(columns1)):
v_j += mutualInformation(i, j, columns1, columns2, pD1,
pD2)
all_j[j] = v_j
column = columns1[0]
n = len(column)
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
mi = mutualInformation(i, j, columns1, columns2, pD1, pD2)
rcwmi[(i, j)] = rowColumnWeighed(mi, i_all[i], all_j[j], n)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(rcwmi[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if rcwmi[(i, j)] != 0.0:
info[(i, j)] = rcwmi[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "cpvn":
Flash('Contact Preferences, Volume Normalized')
cpvn = dict()
score_matrix = mapMatrix("CPVN")
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
average = []
for a, b in zip(columns1[i], columns2[j]):
if a in aa and b in aa:
average.append(
float(matchScore(res1, res2, score_matrix)))
cpvn[(i, j)] = mean(average)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(cpvn[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if cpvn[(i, j)] != 0.0:
info[(i, j)] = cpvn[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "clm":
Flash('Contact PDB-derived Likelihood Matrix')
clm = dict()
score_matrix = mapMatrix("CLM")
for i in range(len(alignment1[0])):
Flash('Column ' + str(i))
for j in range(len(alignment2[0])):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
average = []
for a, b in zip(columns1[i], columns2[j]):
if a in aa and b in aa:
average.append(
float(matchScore(res1, res2, score_matrix)))
clm[(i, j)] = mean(average)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(clm[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if clm[(i, j)] != 0.0:
info[(i, j)] = clm[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "vol":
Flash('Residue-residue Volume Normalized')
vol = dict()
score_matrix = mapMatrix("VOL")
for i in range(len(alignment1[0])):
Flash('Column ' + str(i))
for j in range(len(alignment2[0])):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
average = []
for a, b in zip(columns1[i], columns2[j]):
if a in aa and b in aa:
average.append(
float(matchScore(res1, res2, score_matrix)))
vol[(i, j)] = mean(average)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(vol[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if vol[(i, j)] != 0.0:
info[(i, j)] = vol[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "omes":
Flash('Observed Minus Expected Squared')
omes = dict()
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
omes[(i, j)] = covarianceOMES(columns1[i], columns2[j])
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(omes[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if omes[(i, j)] != 0.0:
info[(i, j)] = omes[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "pearson":
Flash("Pearson's correlation")
pearson = dict()
score_matrix = mapMatrix("MCLACHLAN")
N = len(columns1[0])
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
d_matrix1 = twoDimensionalMatrix(columns1[i], score_matrix)
d_matrix2 = twoDimensionalMatrix(columns2[j], score_matrix)
pearson[(i,
j)] = pearsonsCorrelation(d_matrix1, d_matrix2, N)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(pearson[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if pearson[(i, j)] != 0.0:
info[(i, j)] = pearson[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "spearman":
Flash("Spearman's rank correlation")
score_matrix = mapMatrix("MCLACHLAN")
spearman = dict()
N = len(columns1[0])
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
d_matrix1 = twoDimensionalMatrix(columns1[i], score_matrix)
d_matrix2 = twoDimensionalMatrix(columns2[j], score_matrix)
spearman[(i, j)] = spearmansCorrelation(
d_matrix1, d_matrix2, N)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(spearman[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if spearman[(i, j)] != 0.0:
info[(i, j)] = spearman[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "mcbasc":
Flash('McLachlan Based Substitution Correlation')
mcbasc = dict()
score_matrix = mapMatrix("MCLACHLAN")
N = len(columns1[0])
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
d_matrix1 = twoDimensionalMatrix(columns1[i], score_matrix)
d_matrix2 = twoDimensionalMatrix(columns2[j], score_matrix)
mcbasc[(i, j)] = mcbascCorrelation(d_matrix1, d_matrix2, N)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(mcbasc[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if mcbasc[(i, j)] != 0.0:
info[(i, j)] = mcbasc[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "quartets":
Flash('Quartets')
quartets = dict()
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
quartets[(i, j)] = quartetsCorrelation(
columns1[i], columns2[j])
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(quartets[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if quartets[(i, j)] != 0.0:
info[(i, j)] = quartets[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "sca":
Flash('Statistical Coupling Analysis')
sca = dict()
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
sca[(i, j)] = perturbationSCA(columns1[i], columns2[j], \
j, columns2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(sca[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if sca[(i, j)] != 0.0:
info[(i, j)] = sca[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "elsc":
Flash('Explicit Likelihood of Subset Covariation')
elsc = dict()
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
elsc[(i, j)] = perturbationELSC(columns1[i], columns2[j], \
j, columns2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(elsc[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if elsc[(i, j)] != 0.0:
info[(i, j)] = elsc[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
else:
pass
output = self.dirname + alignment + "_" + coevolution + ".txt"
results = open(output, "w")
for i, j in sorted(info.keys()):
if protein1 != [] and protein2 != []:
print >> results, protein1[i], protein2[j], \
round((info[(i, j)]), 4)
elif protein1 != [] and protein2 == []:
print >> results, protein1[i], protein1[j], \
round((info[(i, j)]), 4)
else:
print >> results, str(i + 1), str(j + 1), \
round((info[(i, j)]), 4)
results.close()
def coevolAnalysis(self, file1, file2, id1, id2,
chain1, chain2, alignment, coevolution):
"Returns a matrix of coevolution scores"
seq = class_sequence(self.file1, self.file2, self.id1, self.id2,
self.chain1, self.chain2, self.parameterfile,
self.dirname)
aln = class_alignment(self.id1, self.id2, self.alignment,
self.parameterfile, self.dirname)
alignment1 = aln.cutAlignment(file1, id1, alignment)
alignment2 = aln.cutAlignment(file2, id2, alignment)
try:
assert len(alignment1) == len(alignment2)
except:
raise StandardError, "Alignments must have the same number of sequences"
protein1 = []
protein2 = []
try:
protein1 = seq.matchResiduePosition(id1, chain1)
protein2 = seq.matchResiduePosition(id2, chain2)
except:
pass
info = dict()
alignment1 = [e for e in alignment1]
columns1 = transpose(alignment1)
alignment2 = [e for e in alignment2]
columns2 = transpose(alignment2)
if coevolution == "mi":
Flash('Mutual Information')
mi = dict()
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
mi[(i, j)] = mutualInformation(i, j, columns1, columns2, pD1, pD2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(mi[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if mi[(i, j)] != 0.0:
info[(i, j)] = mi[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "mie":
Flash('Mutual Information by Pair Entropy')
mie = dict()
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
mie[(i, j)] = miEntropy(i, j, columns1, columns2, pD1, pD2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(mie[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if mie[(i, j)] != 0.0:
info[(i, j)] = mie[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "rcwmi":
Flash('Row and Column Weighed Mutual Information')
rcwmi = dict()
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
i_all = dict()
all_j = dict()
for i in range(len(columns1)):
v_i = 0
for j in range(len(columns2)):
v_i += mutualInformation(i, j, columns1, columns2,
pD1, pD2)
i_all[i] = v_i
for j in range(len(columns2)):
v_j = 0
for i in range(len(columns1)):
v_j += mutualInformation(i, j, columns1, columns2,
pD1, pD2)
all_j[j] = v_j
column = columns1[0]
n = len(column)
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
mi = mutualInformation(i, j, columns1, columns2,
pD1, pD2)
rcwmi[(i, j)] = rowColumnWeighed(mi,
i_all[i], all_j[j], n)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(rcwmi[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if rcwmi[(i, j)] != 0.0:
info[(i, j)] = rcwmi[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "cpvn":
Flash('Contact Preferences, Volume Normalized')
cpvn = dict()
score_matrix = mapMatrix("CPVN")
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
average = []
for a, b in zip(columns1[i], columns2[j]):
if a in aa and b in aa:
average.append(float(matchScore(res1, res2, score_matrix)))
cpvn[(i, j)] = mean(average)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(cpvn[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if cpvn[(i, j)] != 0.0:
info[(i, j)] = cpvn[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "clm":
Flash('Contact PDB-derived Likelihood Matrix')
clm = dict()
score_matrix = mapMatrix("CLM")
for i in range(len(alignment1[0])):
Flash('Column ' + str(i))
for j in range(len(alignment2[0])):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
average = []
for a, b in zip(columns1[i], columns2[j]):
if a in aa and b in aa:
average.append(float(matchScore(res1, res2, score_matrix)))
clm[(i, j)] = mean(average)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(clm[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if clm[(i, j)] != 0.0:
info[(i, j)] = clm[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "vol":
Flash('Residue-residue Volume Normalized')
vol = dict()
score_matrix = mapMatrix("VOL")
for i in range(len(alignment1[0])):
Flash('Column ' + str(i))
for j in range(len(alignment2[0])):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
average = []
for a, b in zip(columns1[i], columns2[j]):
if a in aa and b in aa:
average.append(float(matchScore(res1, res2, score_matrix)))
vol[(i, j)] = mean(average)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(vol[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if vol[(i, j)] != 0.0:
info[(i, j)] = vol[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "omes":
Flash('Observed Minus Expected Squared')
omes = dict()
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
omes[(i, j)] = covarianceOMES(columns1[i], columns2[j])
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(omes[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if omes[(i, j)] != 0.0:
info[(i, j)] = omes[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "pearson":
Flash("Pearson's correlation")
pearson = dict()
score_matrix = mapMatrix("MCLACHLAN")
N = len(columns1[0])
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
d_matrix1 = twoDimensionalMatrix(columns1[i], score_matrix)
d_matrix2 = twoDimensionalMatrix(columns2[j], score_matrix)
pearson[(i, j)] = pearsonsCorrelation(d_matrix1, d_matrix2, N)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(pearson[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if pearson[(i, j)] != 0.0:
info[(i, j)] = pearson[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "spearman":
Flash("Spearman's rank correlation")
score_matrix = mapMatrix("MCLACHLAN")
spearman = dict()
N = len(columns1[0])
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
d_matrix1 = twoDimensionalMatrix(columns1[i], score_matrix)
d_matrix2 = twoDimensionalMatrix(columns2[j], score_matrix)
spearman[(i, j)] = spearmansCorrelation(d_matrix1, d_matrix2, N)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(spearman[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if spearman[(i, j)] != 0.0:
info[(i, j)] = spearman[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "mcbasc":
Flash('McLachlan Based Substitution Correlation')
mcbasc = dict()
score_matrix = mapMatrix("MCLACHLAN")
N = len(columns1[0])
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
d_matrix1 = twoDimensionalMatrix(columns1[i], score_matrix)
d_matrix2 = twoDimensionalMatrix(columns2[j], score_matrix)
mcbasc[(i, j)] = mcbascCorrelation(d_matrix1, d_matrix2, N)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(mcbasc[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if mcbasc[(i, j)] != 0.0:
info[(i, j)] = mcbasc[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "quartets":
Flash('Quartets')
quartets = dict()
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
quartets[(i, j)] = quartetsCorrelation(columns1[i], columns2[j])
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(quartets[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if quartets[(i, j)] != 0.0:
info[(i, j)] = quartets[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "sca":
Flash('Statistical Coupling Analysis')
sca = dict()
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
sca[(i, j)] = perturbationSCA(columns1[i], columns2[j], \
j, columns2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(sca[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if sca[(i, j)] != 0.0:
info[(i, j)] = sca[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
elif coevolution == "elsc":
Flash('Explicit Likelihood of Subset Covariation')
elsc = dict()
for i in range(len(columns1)):
Flash('Column ' + str(i))
for j in range(len(columns2)):
elsc[(i, j)] = perturbationELSC(columns1[i], columns2[j], \
j, columns2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(elsc[(i, j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if elsc[(i, j)] != 0.0:
info[(i, j)] = elsc[(i, j)] * 1.0 / max_val
else:
info[(i, j)] = 0.0
else: pass
output = self.dirname + alignment + "_" + coevolution + ".txt"
results = open(output, "w")
for i, j in sorted(info.keys()):
if protein1 != [] and protein2 != []:
print >> results, protein1[i], protein2[j], \
round((info[(i, j)]), 4)
elif protein1 != [] and protein2 == []:
print >> results, protein1[i], protein1[j], \
round((info[(i, j)]), 4)
else:
print >> results, str(i + 1), str(j + 1), \
round((info[(i, j)]), 4)
results.close()
def coevolAnalysis(self, file1, file2, id1, id2,
chain1, chain2, alignment, coevolution):
"Returns a matrix of coevolution scores"
seq = class_sequence(self.file1, self.file2, self.id1, self.id2,
self.chain1, self.chain2)
aln = class_alignment(self.id1, self.id2, self.alignment)
alignment1 = aln.cutAlignment(file1, id1, alignment)
alignment2 = aln.cutAlignment(file2, id2, alignment)
try:
assert len(alignment1) == len(alignment2)
except:
raise StandardError, "Alignments must have the same number of sequences"
protein1 = []
protein2 = []
try:
protein1 = seq.matchResiduePosition(id1, chain1)
protein2 = seq.matchResiduePosition(id2, chain2)
except:
pass
info = dict()
alignment1 = [e for e in alignment1]
columns1 = transpose(alignment1)
alignment2 = [e for e in alignment2]
columns2 = transpose(alignment2)
if coevolution == "mi":
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
for i in range(len(columns1)):
for j in range(len(columns2)):
info[(i,j)] = mutualInformation(i, j, columns1, columns2, pD1, pD2)
elif coevolution == "mie":
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
for i in range(len(columns1)):
for j in range(len(columns2)):
info[(i,j)] = miEntropy(i, j, columns1, columns2, pD1, pD2)
elif coevolution == "rcwmi":
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
i_all = dict()
all_j = dict()
for i in range(len(columns1)):
v_i = 0
for j in range(len(columns2)):
v_i += mutualInformation(i, j, columns1, columns2,
pD1, pD2)
i_all[i]= v_i
for j in range(len(columns2)):
v_j = 0
for i in range(len(columns1)):
v_j += mutualInformation(i, j, columns1, columns2,
pD1, pD2)
all_j[j]= v_j
column = columns1[0]
n = len(column)
for i in range(len(columns1)):
for j in range(len(columns2)):
mi = mutualInformation(i, j, columns1, columns2,
pD1, pD2)
info[(i,j)] = rowColumnWeighed(mi,
i_all[i], all_j[j], n)
elif coevolution == "cpvnmie":
pD1 = probabilityDict(columns1)
pD2 = probabilityDict(columns2)
for i in range(len(columns1)):
for j in range(len(columns2)):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
mie = miEntropy(i, j, columns1, columns2, pD1, pD2)
info[(i,j)] = contactPreferenceMI(mie, res1, res2)
elif coevolution == "cpvn":
score_matrix = mapMatrix("CPVN")
for i in range(len(columns1)):
for j in range(len(columns2)):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
average = []
for a,b in zip(columns1[i],columns2[j]):
if a in aa and b in aa:
average.append(float(matchScore(res1, res2, score_matrix)))
info[(i,j)] = mean(average)
elif coevolution == "clm":
score_matrix = mapMatrix("CLM")
for i in range(len(alignment1[0])):
for j in range(len(alignment2[0])):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
average = []
for a,b in zip(columns1[i],columns2[j]):
if a in aa and b in aa:
average.append(float(matchScore(res1, res2, score_matrix)))
info[(i,j)] = mean(average)
elif coevolution == "vol":
score_matrix = mapMatrix("VOL")
for i in range(len(alignment1[0])):
for j in range(len(alignment2[0])):
res1 = str(alignment1[0][i])
res2 = str(alignment2[0][j])
average = []
for a,b in zip(columns1[i],columns2[j]):
if a in aa and b in aa:
average.append(float(matchScore(res1, res2, score_matrix)))
info[(i,j)] = mean(average)
elif coevolution == "omes":
omes = dict()
for i in range(len(columns1)):
for j in range(len(columns2)):
omes[(i,j)] = covarianceOMES(columns1[i],columns2[j])
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(omes[(i,j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if omes[(i,j)] != 0.0:
info[(i,j)] = omes[(i,j)] * 1.0 / max_val
else:
info[(i,j)] = 0.0
elif coevolution == "pearson":
score_matrix = mapMatrix("MCLACHLAN")
N = len(columns1[0])
for i in range(len(columns1)):
for j in range(len(columns2)):
d_matrix1 = twoDimensionalMatrix(columns1[i], score_matrix)
d_matrix2 = twoDimensionalMatrix(columns2[j], score_matrix)
info[(i,j)] = pearsonsCorrelation(d_matrix1, d_matrix2, N)
elif coevolution == "spearman":
score_matrix = mapMatrix("MCLACHLAN")
spearman = dict()
N = len(columns1[0])
for i in range(len(columns1)):
for j in range(len(columns2)):
d_matrix1 = twoDimensionalMatrix(columns1[i], score_matrix)
d_matrix2 = twoDimensionalMatrix(columns2[j], score_matrix)
spearman[(i,j)] = spearmansCorrelation(d_matrix1, d_matrix2, N)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(spearman[(i,j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if spearman[(i,j)] != 0.0:
info[(i,j)] = spearman[(i,j)] * 1.0 / max_val
else:
info[(i,j)] = 0.0
elif coevolution == "mcbasc":
score_matrix = mapMatrix("MCLACHLAN")
N = len(columns1[0])
for i in range(len(columns1)):
for j in range(len(columns2)):
d_matrix1 = twoDimensionalMatrix(columns1[i], score_matrix)
d_matrix2 = twoDimensionalMatrix(columns2[j], score_matrix)
info[(i,j)] = mcbascCorrelation(d_matrix1,d_matrix2, N)
elif coevolution == "quartets":
quartets = dict()
for i in range(len(columns1)):
for j in range(len(columns2)):
quartets[(i,j)] = quartetsCorrelation(columns1[i],columns2[j])
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(quartets[(i,j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if quartets[(i,j)] != 0.0:
info[(i,j)] = quartets[(i,j)] * 1.0 / max_val
else:
info[(i,j)] = 0.0
elif coevolution == "sca":
sca = dict()
for i in range(len(columns1)):
for j in range(len(columns2)):
sca[(i,j)] = perturbationSCA(columns1[i],columns2[j],\
j,columns2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(sca[(i,j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if sca[(i,j)] != 0.0:
info[(i,j)] = sca[(i,j)] * 1.0 / max_val
else:
info[(i,j)] = 0.0
elif coevolution == "elsc":
elsc = dict()
for i in range(len(columns1)):
for j in range(len(columns2)):
elsc[(i,j)] = perturbationELSC(columns1[i],columns2[j],\
j,columns2)
max_pos = []
for i in range(len(columns1)):
for j in range(len(columns2)):
max_pos.append(elsc[(i,j)])
max_val = max(max_pos)
for i in range(len(columns1)):
for j in range(len(columns2)):
if elsc[(i,j)] != 0.0:
info[(i,j)] = elsc[(i,j)] * 1.0 / max_val
else:
info[(i,j)] = 0.0
else: pass
output = "./Results/" + alignment + "_" + coevolution + ".txt"
results = open(output, "w")
for i, j in sorted(info.keys()):
if protein1 != [] and protein2 != []:
print >> results, protein1[i], protein2[j], \
round((info[(i, j)]), 4)
elif protein1 != [] and protein2 == []:
print >> results, protein1[i], protein1[j], \
round((info[(i, j)]), 4)
else:
print >> results, str(i+1), str(j+1), \
round((info[(i, j)]), 4)
results.close()
