annotations.py 文件源码

def load_cell_contigs_from_json(json_file, reference_path, group_key, require_high_conf=True):
    """Returns a list of CellContig objects based on annotations in a json.

    The json is assumed to contain a list of AnnotatedContigs (in dict form).
    The contigs are sorted and grouped by group_key and each such group is put
    into a CellContig object.

    group_key must be 'barcode' or 'clonotype'
    """

    assert group_key in set(['barcode', 'clonotype'])
    annotations = load_contig_list_from_json(open(json_file), reference_path)

    cell_contigs = []

    key_func = lambda x: x.__getattribute__(group_key)
    anno_iter = itertools.groupby(sorted(annotations, key=key_func), key=key_func)
    for clonotype_name, contig_annotations in anno_iter:

        contigs = []
        for new_contig in contig_annotations:
            # Note, for consensus contigs is_cell=None
            if new_contig.is_cell is not False \
               and (new_contig.high_confidence or not require_high_conf):
                contigs.append(new_contig)

        if len(contigs) > 0:
            cell_contigs.append(CellContigs(clonotype_name, contigs))

    return cell_contigs