Exemple #1
0
def make_basecall_input_multi(fast5_files, section='template', window=[-1, 0, 1], trim=10, min_len=1000, max_len=9000,
    event_detect=True, ed_params={'window_lengths':[3, 6], 'thresholds':[1.4, 1.1], 'peak_height':0.2}, sloika_model=False):
    """Like the above, but doesn't yields directly events. The point here is to
    be fully consistent with the currennt interface but allow use of the python
    library
    """
    for f in fast5_files:
        with Fast5(f) as fh:
            if event_detect:
                # These parameters make no sense to me, but hey-ho
                # TODO: expose to user
                events = minknow_event_detect(
                    fh.get_read(raw=True), fh.sample_rate, **ed_params
                )
            else:
                events = fh.get_read()
            events, _ = segment(events, section=section) 
        try:
            X = events_to_features(events, window=window, sloika_model=sloika_model)
        except TypeError:
            continue
        try:
            X = X[trim:-trim]
            events = events[trim:-trim]
        except:
            continue
        else:
            if len(X) < min_len or len(X) > max_len:
                continue
        yield f, X, events
Exemple #2
0
def basecall_file(fname=None, event_detect=True):
    """Read event data from file and print scrappie basecall.

    :param fname: filename to read data from (if not given assumed
        to be given on command line.
    :param event_detect: do event detection?

    :returns: tuple (basecall score, sequence).
    """
    is_main = False
    if fname is None: #called as entrypoint
        fname = sys.argv[1]
        is_main = True

    # magic numbers
    ed_params = {
        'window_lengths':[4, 8],
        'thresholds': [1.5, 9.0],
        'peak_height': 0.2,
    }

    with Fast5(fname) as fh:
        if event_detect:
            events = minknow_event_detect(
                fh.get_read(raw=True), fh.sample_rate, **ed_params
            )
        else:
            events = fh.get_read()
    events, _ = segment(events, section='template') 

    results = basecall_events(events)
    if results is None:
        return None
    if is_main:
        print("{} score={}\n{}".format(fname, *results))
    else:
        return results
Exemple #3
0
def basecall_file(fname=None, event_detect=True):
    """Read event data from file and print scrappie basecall.

    :param fname: filename to read data from (if not given assumed
        to be given on command line.
    :param event_detect: do event detection?

    :returns: tuple (basecall score, sequence).
    """
    is_main = False
    if fname is None:  #called as entrypoint
        fname = sys.argv[1]
        is_main = True

    # magic numbers
    ed_params = {
        'window_lengths': [4, 8],
        'thresholds': [1.5, 9.0],
        'peak_height': 0.2,
    }

    with Fast5(fname) as fh:
        if event_detect:
            events = minknow_event_detect(fh.get_read(raw=True),
                                          fh.sample_rate, **ed_params)
        else:
            events = fh.get_read()
    events, _ = segment(events, section='template')

    results = basecall_events(events)
    if results is None:
        return None
    if is_main:
        print("{} score={}\n{}".format(fname, *results))
    else:
        return results