parsimony_class.py

from nwck_class_3 import Newick_Tree

class Parsimony:
    def __init__(self,tree,sequences,alphabet,length):
        self.sequences=sequences
        self.seq=dict()
        self.costV=dict()
        self.cost=dict()
        self.tree=tree
        self.alphabet=alphabet
        self.N=length
        
        for taxon in tree.labels.keys():
            self.seq[tree.labels[taxon]]=self.sequences[taxon]
            for i in range(self.N):
                self.costV[(tree.labels[taxon],i,sequences[taxon][i])]=0
                self.cost[(tree.labels[taxon],i)]=0
        self.sankoff()
        
    def sankoff(self,base=None):
        if not base:
            base=self.tree.root
        if len(self.tree.dtree[base])==0:
            return
        for v in self.tree.dtree[base]:
            self.sankoff(v)
        for i in range(self.N):   
            for c in self.alphabet:
                m=0
                for v in self.tree.dtree[base]:                     
                    if (v,i,c) in self.costV:
                        m=m+min(self.costV[(v,i,c)],self.cost[(v,i)]+1)
                    else:
                        m=m+self.cost[(v,i)]+1
                self.costV[(base,i,c)]=m
            self.cost[(base,i)]=min([self.costV[(base,i,c)] for c in self.alphabet])
        return
        
    def score(self):
        return sum([self.costV[(self.tree.root,i,self.sequences[self.tree.taxa[self.tree.node_to_taxon[self.tree.root]]][i])] for i in range(self.N)])