ORFsClustering.py 文件源码

def from_orfs_files(cls, seq_path, paths_path, graph_path):
        """
        Create object from graph, ORFs files
        :param seq_path: path to .fasta file with ORFs sequences
        :param paths_path: path to .path file with ORF's paths
        :param graph_path: path to graph
        :return: LinkageCluster
        """
        from Bio import SeqIO
        orfs = collections.OrderedDict()
        with open(seq_path) as seq_file, open(paths_path) as path_file:
            paths_dict = {}
            orfid = ''
            for ind, line in enumerate(path_file.read().split('\n')):
                if ind % 2 == 0:
                    try:
                        orfid = re.findall(r'(ORF_\d+)', line)[0]
                    except:
                        continue
                else:
                    paths_dict[orfid] = [int(re.findall(r'^(\d+),', line)[0])] + re.findall('(\d+)([+-]),', line) +\
                                               [int(re.findall(r',(\d+)$', line)[0])]
            for rec in SeqIO.parse(seq_file, 'fasta'):
                orfs[str(rec.seq)] = (paths_dict[rec.id], rec.id)
        return cls(GFAgraph().load_graph(graph_path), orfs)