vFam_HmmSearch_parse.py 文件源码

def vfam_to_krona(self, vfam_file):
        vfam_dic = defaultdict(int)
        families_dic = {}
        genera_dic = {}
        with open(vfam_file, 'r') as vfam_file:
            vfam_file.readline()  # get rid of the header
            for line in vfam_file:
                splitted_line = line.split('\t')

                vfam = splitted_line[1]
                vfam_dic[vfam] += 1

                families = ast.literal_eval(splitted_line[3])  # safe eval of dict
                families_dic[vfam] = families
                genera = ast.literal_eval(splitted_line[4])
                genera_dic[vfam] = genera

        with open(self.krona_in, 'w') as o:
            for vfam, n_reads in vfam_dic.items():
                fam_total = sum(families_dic[vfam].values())
                for fam, fam_prop in families_dic[vfam].items():
                    gen_total = sum(genera_dic[vfam].values())
                    for genera, gen_prop in genera_dic[vfam].items():
                        n = (n_reads * (fam_prop / fam_total)) * (gen_prop / gen_total)
                        o.write('%.3f\t%s\t%s\t%s\n' % (n, fam, vfam, genera))