parsimony.py

from nwck_class_3 import Newick_Tree
from collections import defaultdict
from my_iterators import Tree
from parsimony_class import Parsimony
  
sequences=dict()  
with open("Lewis/angio35Sequences6.txt","rU") as f:
    for line in f:
        taxon,sequence=line.split()
        sequences[taxon]=sequence


bad=[]  
      
for x in sequences.keys():
    for i,y in enumerate(sequences[x]):
        if y.upper() not in ['C','A','G','T'] and i not in bad:
            bad.append(i)
#good=[]            
#
#k=sequences.keys()[0]
#t=sequences[k]
#for i in range(len(t)):
#    for u in sequences.keys():
#        if sequences[u][i]!=t[i]:
#            good.append(i)
            
for x in sequences.keys():
    fixed=''
    for i,y in enumerate(sequences[x]):
        if i  not in bad:
            fixed+=y.upper()
    sequences[x]=fixed

spectrum=[]
taxa=sequences.keys()
N=min([len(x) for x in sequences.values()])
for x in Tree(taxa):
    m=Parsimony(Newick_Tree(x),sequences,['A','C','G','T'],N)
    print x,m.score()
    spectrum.append(m.score())
    
print sorted(spectrum)
                


#x=z[1:]

#for t in trees(x):
#    print '(%s)%s;'%(lists_to_newick(t),z[0])