def detect_correlations(self,site):
"""Does the MLP exhibit any correlations of the type
demonstrated by Man & Stormo (NAR 2001)?"""
bases = "ATCG" # follow ordering defined in Recognizer class
w = len(site)
all_dependent = True
dependencies = 0
discrepancy_dict = {}
for (i,j) in choose2(range(w)):
print (i,j)
base1 = site[i]
base2 = site[j]
base_index1 = Recognizer.base_index[base1]
base_index2 = Recognizer.base_index[base2]
default_energy = self.binding_energy(site)
alt_sites = [[subst(subst(site,b1,i),b2,j)
for b2 in bases]
for b1 in bases]
alt_energies = mmap(self.binding_energy,alt_sites)
best_energy = min(map(min,alt_energies))
norm_energies = mmap(lambda x:x/best_energy,alt_energies)
norm_default_energy = default_energy/best_energy
independent = all((sorted_indices(alt_energy_row) ==
sorted_indices(alt_energies[0]))
for alt_energy_row in alt_energies)
#independent = False
if independent:
all_dependent = False
else:
dependencies += 1
###Print motif
print (i,j)
print " "
for c2 in bases:
print " ",c2,
print
for b_i1,c1 in enumerate(bases):
print c1,
for b_i2 in range(4):
print "%1.4f" % norm_energies[b_i1][b_i2],
print
###Finish printing motif
base_1_mutant_energies = [(b,row[base_index2])
for (b,row) in zip(bases,norm_energies)]
print "base 1 mutants:",base_1_mutant_energies
best_responses = [min(zip(bases,row),
key=lambda (b,x):abs(norm_default_energy - x))
for row in norm_energies]
print "best responses:",best_responses
discrepancy = max(zip(base_1_mutant_energies, best_responses),
key = lambda ((b1,x),(b2,y)): abs(x-y))
print "actual bases:",base1,base2,(base_index1,base_index2)
print "norm default energy:",norm_default_energy
(b1,v1),(b2,v2) = discrepancy
print "discrepancy:",b1,b2,v1,v2,v2-v1
discrepancy_dict[(i,j)] = v2-v1
print "all dependent:",all_dependent
print "# dependencies:", dependencies, "out of:",len(choose2(range(w)))
return discrepancy_dict
def detect_correlations2(self,site):
"""Does the MLP exhibit any correlations of the type
demonstrated by Man & Stormo (NAR 2001)?"""
bases = "ATCG" # follow ordering defined in Recognizer class
w = len(site)
all_dependent = True
dependencies = 0
discrepancy_dict = {}
for (i,j) in choose2(range(w)):
print (i,j)
base1 = site[i]
base2 = site[j]
base_index1 = Recognizer.base_index[base1]
base_index2 = Recognizer.base_index[base2]
default_energy = self.binding_energy(site)
alt_sites = [[subst(subst(site,b1,i),b2,j)
for b2 in bases]
for b1 in bases]
alt_energies = mmap(self.binding_energy,alt_sites)
best_energy = min(map(min,alt_energies))
norm_energies = mmap(lambda x:x/best_energy,alt_energies)
norm_default_energy = default_energy/best_energy
independent = all((sorted_indices(alt_energy_row) ==
sorted_indices(alt_energies[0]))
for alt_energy_row in alt_energies)
#independent = False
if independent:
all_dependent = False
else:
dependencies += 1
###Print motif
print (i,j)
print " "
for c2 in bases:
print " ",c2,
print
for b_i1,c1 in enumerate(bases):
print c1,
for b_i2 in range(4):
print "%1.4f" % norm_energies[b_i1][b_i2],
print
print "actual bases:",base1,base2,(base_index1,base_index2)
###Finish printing motif
for base,row in zip(bases,norm_energies):
if (norm_energies[base_index1][base_index2] > row[base_index2] and
(all(norm_energies[base_index1][b] < row[b]
for b in range(4) if b != base_index1))):
print "discrepancy:", (norm_energies[base_index1][base_index2] -
row[base_index2])
print "consider row:",base,row
tr_norm_energies = transpose(norm_energies)
for base,col in zip(bases,tr_norm_energies):
if (norm_energies[base_index1][base_index2] > col[base_index1]
and
(all(norm_energies[b][base_index2] < col[b]
for b in range(4) if b != base_index1))):
print "discrepancy:",(norm_energies[base_index1][base_index2] -
col[base_index1])
print "consider col:",base,col
print "all dependent:",all_dependent
print "# dependencies:", dependencies, "out of:",len(choose2(range(w)))
return discrepancy_dict
