lib5c.algorithms.donut_filters module

lib5c.algorithms.donut_filters.apply_filter(obs_matrix, exp_matrix, footprint, max_percent=0.2, min_exp=0.1)[source]

Computes a corrected expected value by applying a footprint to observed and expected matrix.

Parameters:
  • exp_matrix (obs_matrix,) – The square, symmetric matrices of observed and expected counts, respectively.
  • footprint (np.ndarray) – The footprint to convolve. Should contain 1’s at positions included in the footprint and 0’s everywhere else.
  • max_percent (float) – If the proportion of nan’s in the footprint for a pixel is greater than this value, the corrected expected at that point will be set to nan.
  • min_exp (float) – If the sum of entries in exp_matrix under the footprint for a particular pixel is less than this value, set the output at this pixel to nan to avoid numerical instability related to division by small numbers.
Returns:

The corrected expected value.
Return type:

np.ndarray
lib5c.algorithms.donut_filters.donut_filt(obs_matrix, exp_matrix, p, w, max_percent=0.2, min_exp=0.1)[source]

Computes the full donut filter.

Parameters:
  • exp_matrix (obs_matrix,) – The square, symmetric matrices of observed and expected counts, respectively.
  • w (p,) – The inner and outer radii of the donut, respectively.
  • max_percent (float) – If the proportion of nan’s in the footprint for a pixel is greater than this value, the corrected expected at that point will be set to nan.
  • min_exp (float) – If the sum of entries in exp_matrix under the footprint for a particular pixel is less than this value, set the output at this pixel to nan to avoid numerical instability related to division by small numbers.
Returns:

The corrected expected value.
Return type:

np.ndarray
lib5c.algorithms.donut_filters.lower_left_filt(obs_matrix, exp_matrix, p, w, max_percent=0.2, min_exp=0.1)[source]

Computes the lower left donut filter.

Parameters:
  • exp_matrix (obs_matrix,) – The square, symmetric matrices of observed and expected counts, respectively.
  • w (p,) – The inner and outer radii of the donut, respectively.
  • max_percent (float) – If the proportion of nan’s in the footprint for a pixel is greater than this value, the corrected expected at that point will be set to nan.
  • min_exp (float) – If the sum of entries in exp_matrix under the footprint for a particular pixel is less than this value, set the output at this pixel to nan to avoid numerical instability related to division by small numbers.
Returns:

The corrected expected value.
Return type:

np.ndarray

Examples

>>> import numpy as np
>>> from lib5c.algorithms.donut_filters import lower_left_filt
>>> from lib5c.algorithms.expected import empirical_binned
>>> from lib5c.algorithms.expected import make_expected_matrix_from_list
>>> obs = np.array([[10,  4,  1],
...                 [ 4,  8,  6],
...                 [ 1,  6, 12]]).astype(float)
>>> exp = make_expected_matrix_from_list(
...     empirical_binned(obs, log_transform=False))
>>> exp
array([[10.,  5.,  1.],
       [ 5., 10.,  5.],
       [ 1.,  5., 10.]])
>>> lower_left_filt(obs, exp, 0, 1, max_percent=0.0, min_exp=0.0)
array([[ nan,  4. ,  0.8],
       [ 4. , 10. ,  6. ],
       [ 0.8,  6. ,  nan]])