annotations.py 文件源码

def get_unannotated_intervals(self):
        """ Return a list of Annotation objects corresponding to unannotated regions on the contig """

        unannotated_intervals = []
        annotation_indicator = np.zeros(len(self.sequence))

        for annotation in self.annotations:
            annotation_indicator[annotation.contig_match_start:annotation.contig_match_end] = 1

        interval_start = 0

        for annotated, region_iter in itertools.groupby(annotation_indicator, lambda x: x == 1):
            region = list(region_iter)

            if not annotated:
                feature = vdj_reference.create_dummy_feature(display_name='UNANNOTATED',
                                                             region_type='UNANNOTATED',
                                                             sequence=None)
                unannotated_intervals.append(Annotation(feature=feature,
                                                        cigar=None,
                                                        score=0,
                                                        annotation_length=len(region),
                                                        annotation_match_start=0,
                                                        annotation_match_end=len(region),
                                                        contig_match_start=interval_start,
                                                        contig_match_end=interval_start + len(region),
                                                        mismatches=[],
                                                    ))

            interval_start += len(region)

        return unannotated_intervals