neuron_morphology.features.layer.layer_histogram
¶
Module Contents¶
Classes¶
LayerHistogram |
The results of calculating a within-layer depth histogram of points |
EarthMoversDistanceInterpretation |
Describes how to understand an earth mover’s distance result. This is |
EarthMoversDistanceResult |
The result of comparing two histograms using earth mover’s distance |
Functions¶
ensure_tuple (inputs: Any, item_type: Type, if_none: Union[str, Tuple] = ‘raise’) → Tuple |
Try to smartly coerce inputs to a tuple. |
ensure_node_types (node_types) |
Make sure the argued node types are a tuple |
ensure_layers (layers) |
Make sure the argued layer array is a tuple |
earth_movers_distance (data: Data, node_types: Sequence[int], node_types_to_compare: Sequence[int], bin_size: float = 5) → Dict[str, EarthMoversDistanceResult] |
Calculate the earth mover’s distance between normalized histograms of |
histogram_earth_movers_distance (from_hist: np.ndarray, to_hist: np.ndarray) → EarthMoversDistanceResult |
Calculate the earth mover’s distance between to histograms, normalizing |
normalized_depth_histogram (data: Data, node_types: Optional[Sequence[int]] = None, bin_size=5.0) → Dict[str, LayerHistogram] |
Calculates for each cortical layer a histogram of node depths within |
normalized_depth_histograms_across_layers (data: Data, point_types: Optional[Tuple[int]] = None, only_layers: Optional[Tuple[str]] = None, bin_size=5.0) → Dict[str, LayerHistogram] |
A helper function for running cortical depth histograms across multiple |
normalized_depth_histogram_within_layer (point_depths: np.ndarray, local_layer_pia_side_depths: np.ndarray, local_layer_wm_side_depths: np.ndarray, reference_layer_depths: ReferenceLayerDepths, bin_size: float) → np.ndarray |
Calculates a histogram of node depths within a single (cortical) layer. |
-
neuron_morphology.features.layer.layer_histogram.
ensure_tuple
(inputs: Any, item_type: Type, if_none: Union[str, Tuple] = 'raise') → Tuple¶ Try to smartly coerce inputs to a tuple.
Parameters: - inputs : the data to be coerced
- item_type : which type do/should the elements of the tuple have?
- if_none : if the inputs are none, return this value. If the value is
“raise”, instead raise an exception
Returns: - the coerced inputs
-
neuron_morphology.features.layer.layer_histogram.
ensure_node_types
(node_types)¶ Make sure the argued node types are a tuple
-
neuron_morphology.features.layer.layer_histogram.
ensure_layers
(layers)¶ Make sure the argued layer array is a tuple
-
class
neuron_morphology.features.layer.layer_histogram.
LayerHistogram
¶ Bases:
typing.NamedTuple
The results of calculating a within-layer depth histogram of points within some cortical layer.
-
counts
:np.ndarray¶
-
bin_edges
:np.ndarray¶
-
-
class
neuron_morphology.features.layer.layer_histogram.
EarthMoversDistanceInterpretation
¶ Bases:
enum.Enum
Describes how to understand an earth mover’s distance result. This is useful in the case that one or both histograms are all 0.
-
BothPresent
= 0¶
-
OneEmpty
= 1¶
-
BothEmpty
= 2¶
-
-
class
neuron_morphology.features.layer.layer_histogram.
EarthMoversDistanceResult
¶ Bases:
typing.NamedTuple
The result of comparing two histograms using earth mover’s distance
-
result
:float¶
-
interpretation
:EarthMoversDistanceInterpretation¶
-
to_dict_human_readable
(self)¶
-
-
neuron_morphology.features.layer.layer_histogram.
earth_movers_distance
(data: Data, node_types: Sequence[int], node_types_to_compare: Sequence[int], bin_size: float = 5) → Dict[str, EarthMoversDistanceResult]¶ Calculate the earth mover’s distance between normalized histograms of node depths within cortical layers. Calculates one distance for each layer.
Parameters: - data : Must be endowed with layered_point_depths and reference_layer_depths.
The morphology is not actually used directly.
- node_types : Defines one set of points whose histograms to compare.
- node_types_to_compare : Defines the other set of points
- bin_size : the size of each depth bin. Default is appropriate if the units
are microns.
Returns: - A mapping from layers to distances between histograms within those layers.
-
neuron_morphology.features.layer.layer_histogram.
histogram_earth_movers_distance
(from_hist: np.ndarray, to_hist: np.ndarray) → EarthMoversDistanceResult¶ Calculate the earth mover’s distance between to histograms, normalizing each. If one histogram is empty, return the sum of the other and a flag. If both are empty, return 0 a and a flag.
Parameters: - from_hist : distance is calculated between (the normalized form of) this
histogram and to_hist. The result is symmetric.
- to_hist : distance is calculated between (the normalized form of) this
histogram and from_hist
Returns: - The distance between input histograms, along with an enum indicating
- whether one or both of the histograms was all 0.
-
neuron_morphology.features.layer.layer_histogram.
normalized_depth_histogram
(data: Data, node_types: Optional[Sequence[int]] = None, bin_size=5.0) → Dict[str, LayerHistogram]¶ Calculates for each cortical layer a histogram of node depths within that layer.
Parameters: - data : Must have the following attributes:
- reference_layer_depths : A dictionary mapping layer names (str) to
ReferenceLayerDepths objects describing the average pia and white- matter side depths of this each layer.
- layered_point_depths : A LayeredPointDepths defining for each point a
depth from pia. See LayeredPointDepths for more information.
- node_types : for which to calculate the histograms
- bin_size : the size of each depth bin. Default is appropriate if the units
are microns.
-
neuron_morphology.features.layer.layer_histogram.
normalized_depth_histograms_across_layers
(data: Data, point_types: Optional[Tuple[int]] = None, only_layers: Optional[Tuple[str]] = None, bin_size=5.0) → Dict[str, LayerHistogram]¶ A helper function for running cortical depth histograms across multiple layers.
Parameters: - data : must have reference_layer_depths and layered_point_depths
- point_types : calculate histograms for points labeled with these types
- only_layers : exclude other layers from this calculation
- bin_size : the size of each depth bin. Default is appropriate if the units
are microns.
-
neuron_morphology.features.layer.layer_histogram.
normalized_depth_histogram_within_layer
(point_depths: np.ndarray, local_layer_pia_side_depths: np.ndarray, local_layer_wm_side_depths: np.ndarray, reference_layer_depths: ReferenceLayerDepths, bin_size: float) → np.ndarray¶ Calculates a histogram of node depths within a single (cortical) layer. Uses reference information about layer boundaries to normalize these depths for cross-reconstruction comparison.
Parameters: - depths : Each item corresponds to a point of interest (such as a node
in a morphological reconstruction). Values are the depths of these points of interest from the pia surface.
- local_layer_pia_side_depths : Each item corresponds to a point of interest.
Values are the depth of the intersection point between a path of steepest descent from the pia surface to the point of interest and the upper surface of the layer.
- local_layer_wm_side_depths : Each item corresponds to a point of interest.
Values are the depth of the intersection point between the layer’s lower boundary and the path described above.
- reference_layer_depths : Used to provide normalized depths suitable
for comparison across reconstructions. Should provide a generic equivalent of local layer depths for a population or reference space.
- bin_size : The width of each bin, in terms of depths from pia in the
reference space. Provide only one of bin_edges or bin_size.
Returns: - A numpy array listing for each depth bin the number of nodes falling within
that bin.
Notes
This function relies on the notion of a steepest descent path through cortex, but is agnostic to the method used to obtain such a path and to features of the path (e.g. whether it is allowed to curve). Rather the caller must ensure that all depths have been calculated according to a consistent scheme.