import numpy as np
from ..tbl.table import Table
[docs]class CutOffPixelQuantile(Table):
"""Select table rows by the intensity count quantile.
Noise distribution seems to be Gaussian distribution. However, the
image includes signals and results in gamma-like distribution. This
function uses Median + factor * (Q2 - Q1) as intensity threshold
instead of Mean + factor * STD to avoid using signal-biased STD.
.. caution::
The threshold is calculated from all rows entered into the
:meth:`process`. You have to split the required data into appropriate
depths.
Args:
reqs[0] (Table): Table including intensity values.
param["calc_col"] (str): Column name for calculating the median.
param["cut_factor"] (float): Cutoff factor above the median value.
param["ignore_zero"] (bool, optional): Whether zero values are ignored
from the intensity.
param["split_depth"] (int): File split depth number.
Returns:
Table: Selected Table
"""
[docs] def set_info(self, param={}):
"""Copy info from reqs[0] and add params.
"""
self.info.copy_req(0)
self.info.add_param(
"cut_factor", param["cut_factor"], "num", "Quantile factor")
self.info.add_param(
"calc_col", param["calc_col"], "str",
"Column name for calculating median")
if ("ignore_zero", True) in param.items():
self.info.add_param(
"ignore_zero", param["ignore_zero"], "bool",
"Whether zero values are ignored from the intensity")
self.info.set_split_depth(param["split_depth"])
[docs] @ staticmethod
def process(reqs, param):
"""Select table rows by the intensity count quantile.
Args:
reqs[0] (pandas.DataFrame): Table including intensity values.
param["calc_col"] (str): Column name for calculating the median.
param["cut_factor"] (float): Cutoff factor above the median value.
param["ignore_zero"] (bool, optional): Whether zero values are
ignored from the intensity.
Returns:
pandas.DataFrame: Selected table
"""
df = reqs[0].copy()
intensity = df[param["calc_col"]].values
if ("ignore_zero", True) in param.items():
intensity = intensity[np.nonzero(intensity)]
pct = np.percentile(intensity, q=[25, 50])
dif_q = pct[1] - pct[0]
threshold = pct[1] + dif_q * param["cut_factor"]
return df[df[param["calc_col"]] > threshold]