GO的is-a关系树 输入为原始obo文件 结果约56万行

go_is-a_tree.py

import sys
import re

id="id"
ns="namespace"
ia="is-a"
nps="node-ptrs"

def get_terms_namespaces(go_file):
    terms={}
    namespaces={}
    for info in open(go_file,'r').read().split("[Term]")[1:]:
        term={}
        lines=info.splitlines()
        is_a=[]
        for line in lines:
            if re.match(r"^id: GO",line):
                term[id]=line.split()[1]
            if re.match(r"^namespace",line):
                term[ns]=line.split()[1]
            if re.match(r"^is_a: GO",line):
                is_a.append(line.split()[1])
        term[ia]=is_a
        term[nps]=[]
        terms[term[id]]=term
        namespaces[term[ns]]=1
    return terms,namespaces.keys()

class node(object):
    def __init__(self,parent,name):
        self.parent=parent
        self.name=name
        self.childs=[]
    def __str__(self,level=0):
        s=("[--"*level)+self.name+"\n"
        if len(self.childs)>1:
            for child in self.childs:
                s=s+child.__str__(level+1)
        return s
    def add_child(self,child_id):
        child=node(self,child_id)
        self.childs.append(child)
        return child

def build_tree(terms,namespaces):
    root=node(None,"ROOT")
    nss_dic={}
    for single_ns in namespaces:
        nss_dic[single_ns]=root.add_child(single_ns)
    
    def add_term(term):
        # print "getID: "+term[id]+"/len:"+str(len(term[nps]))
        if len(term[nps])>0:
            # print "IsIn: "+term[id]
            return term[nps]
        if len(term[ia])<1:
            parent=nss_dic[term[ns]]
            np=parent.add_child(term[id])
            # print "NameAs["+term[ns]+"]: "+term[id]
            return [np]
        nodes=[]
        for parent_id in term[ia]:
            parent=terms[parent_id]
            parent[nps]=add_term(parent)
            for parent_node in parent[nps]:
                if term[id] in [child.name for child in parent_node.childs]:
                    continue
                # print "takeBeweeteen: "+parent_id+" <-> "+term[id]
                nodes.append(parent_node.add_child(term[id]))
        return nodes

    for term_id in terms.keys():
        term_p=terms[term_id]
        term_p[nps]=add_term(term_p)
    return root

if __name__=="__main__":
    f=sys.argv[1]
    terms,nss=get_terms_namespaces(f)
    tree=build_tree(terms,nss)
    print tree