def remove_outliers(self, configuration):
if self._mixed_windows is None:
raise Error("Mixed windows have not been computed")
# compute the statistis
wga_rds = []
no_wga_rds = []
for window in self._mixed_windows:
if not window.is_n_window():
wga_rds.extend(window.get_rd_counts(name=WindowType.WGA))
no_wga_rds.extend(window.get_rd_counts(name=WindowType.NO_WGA))
wga_statistics = compute_statistic(data=wga_rds, statistics="all")
no_wga_statistics = compute_statistic(data=no_wga_rds,
statistics="all")
config = configuration["outlier_remove"]["config"]
config["statistics"] = {
WindowType.NO_WGA: no_wga_statistics,
WindowType.WGA: wga_statistics
}
self._mixed_windows = \
remove_outliers(windows=self._mixed_windows,
removemethod=configuration["outlier_remove"]["name"],
config=config)
def remove_outliers(self, configuration):
if self._mixed_windows is None:
raise Error("Mixed windows have not been computed")
# compute the statistics
wga_means = array.array('d')
no_wga_means = array.array('d')
for window in self._mixed_windows:
if not window.is_gap_window():
wga_means.append(window.get_rd_statistic(statistics="mean",
name=WindowType.WGA))
no_wga_means.append(window.get_rd_statistic(statistics="mean",
name=WindowType.NO_WGA))
if len(wga_means) == 0 or len(no_wga_means) == 0:
print("{0} Cannot remove outliers for region. "
"Empty RD list detected".format(WARNING))
return
wga_statistics = compute_statistic(data=wga_means,
statistics="all")
no_wga_statistics = compute_statistic(data=no_wga_means,
statistics="all")
config = configuration["outlier_remove"]["config"]
config["statistics"] = {WindowType.NO_WGA: no_wga_statistics,
WindowType.WGA: wga_statistics}
self._mixed_windows = \
remove_outliers(windows=self._mixed_windows,
removemethod=configuration["outlier_remove"]["name"],
config=config)
def zscore_statistic(data, null, **kwargs):
statistic = compute_statistic(data=data, statistics=null.name)
# compute the variance
var = compute_statistic(data=data, statistics="var")
score = (statistic - null.value) / (np.sqrt(var))
if "alternative" in kwargs:
direction = kwargs["alternative"].direction
if direction == ">" or direction == ">=":
prob = 1.0 - st.norm.cdf(score)
return prob, statistic
elif direction == "<" or direction == "<=":
prob = st.norm.cdf(score)
return prob, statistic
else:
prob = 2.0 * (1.0 - st.norm.cdf(np.abs(score)))
return prob, statistic
else:
# assume two-sided by default
prob = 2.0 * (1.0 - st.norm.cdf(np.fabs(score)))
return prob, statistic
def get_rd_stats(self, statistics="all"):
"""
Returns a statistical summary as a dictionary
of the read depth variable in the window
:param statistics:
:return:
"""
# accumulate RD as an array and use numpy
rd_data = [item.read_depth for item in self._observations]
from preprocess_utils import compute_statistic
return compute_statistic(data=rd_data,
statistics=statistics)
def get_statistics(self, statistic, window_type, **kwargs):
if window_type == WindowType.BOTH:
for index in self._indexes:
window = self._windows[index]
statistic1, statistic2 = \
window.get_rd_stats(statistics=statistic)
return statistic1, statistic2
else:
wga_windows = [
window.get_window(window_type) for window in self._windows
]
window_data = flat_windows_rd_from_indexes(indexes=self._indexes,
windows=wga_windows)
return compute_statistic(data=window_data, statistics=statistic)