Beispiel #1
0
    def __init__(self, gtf_file, landmarks=ALL_LANDMARKS, use_strand=True, **kwargs):
        super(DistToClosestLandmarkExtractor, self).__init__(gtf_file, **kwargs)
        self._gtf_file = gtf_file
        self.landmarks = extract_landmarks(gtf_file, landmarks=landmarks)
        self.columns = landmarks  # column names. Reqired for concating distances into array
        self.use_strand = use_strand

        # set index to chromosome and strand - faster access
        self.landmarks = {k: v.set_index(["seqname", "strand"])
                          for k, v in six.iteritems(self.landmarks)}
Beispiel #2
0
def test_extract_landmarks(gtf):
    """Run for all the extractors
    """
    ldm = extract_landmarks(gtf)
    assert isinstance(ldm, dict)