fname = "".join([str(n),"_Mchains.p"])
Mchains = pickle.load(open(fname,"r"))
print "fitting model to test set..."
a = 0
for line in fileinput.input():
l = line.strip("\n")
l = l.split("\t")
l[0] = l[0].split("_")[0]
if len(l[1])<=n:
continue
target = l[0]
answer = np.array(Mchains.keys())
estimates = np.zeros( (len(Mchains.keys()),1))
temp = create_chain(len(ix),1)
l[1] = l[1].replace("N",randomN())
temp = process_read(l[1],temp,n,ix)
temp = temp/float(sum(temp))
for k,val in enumerate(Mchains.keys()):
prob = euclid(temp,Mchains[val])
estimates[k] = np.log2(prob)
best = np.argmin(estimates)
print answer[best], target
if answer[best] == target:
a+=1
print "total of ",a, "fitted properly"
'''create a dictionnary containing all markov chains (order 6) for each chromosome'''
Mchains = {}
probchr = {}
a = 0
for line in fileinput.input():
l = line.strip("\n")
l = l.split("\t")
l[0] = l[0].split("_")[0]
if len(l[1])<=n:
continue
if not ( l[0] in Mchains.keys()):
Mchains[l[0]] = create_chain(len(ix),1)
probchr[l[0]] = 0
temp = create_chain(len(ix),1)
l[1] = l[1].replace("N",randomN())
Mchains[l[0]]+=process_read(l[1],temp,n,ix)
probchr[l[0]]+=1
sum_prob = float(np.sum(probchr.values()))
#calculating Poisson distribution parameters (mu) for each Markov chain
for k in Mchains.keys():
Mchains[k] = Mchains[k]/float(sum(Mchains[k]))
probchr[k] = probchr[k]/sum_prob
''' For each chromosome, the occurences are converted to
probability tables.
For each chromosome, a |1-distance|**2 is calculated as a metric'''
fname = "".join([str(n),"_Mchains.p"])
pickle.dump(Mchains,open((fname),"w"))
