BrainLine¶
Data Helper Functions¶
-
brainlit.BrainLine.download_subvolumes(data_dir: str, brain_id: str, layer_names: list, dataset_to_save: str, data_file: str)[source]¶ Download subvolumes around a set of manually marked points for validation of machine learning model.
- Parameters
data_dir (str) -- Path to directory where subvolumes will be saved.
brain_id (str) -- Brain ID key in brain2paths dictionary from soma_data or axon_data/
layer_names (list) -- List of precomputed layer names associated with the brain_id, ordered by primary signal channel (e.g. antibody), background channel, and secondary signal channel (e.g. endogenous fluorescence).
dataset_to_save (str) -- val or train - specifies which set of subvolumes should be downloaded, if applicable.
data_file (str) -- path to json file with data information.
- Raises
ValueError -- If object_type is not soma or axon
Apply Ilastik and Validate Models¶
-
class
brainlit.BrainLine.ApplyIlastik(ilastik_path: str, project_path: str, brains_path: str, brains: list)[source]¶ Applies ilastik to subvolumes for the purpose of validating machine learning algorithms.
- Parameters
-
brains_path¶ Path to directory that contains brain samples subdirectories.
- Type
Path
-
brainlit.BrainLine.plot_results(data_dir: str, brain_ids: list, object_type: str, positive_channel: int, dset: str = 'val', doubles: list = [], show_plot: bool = True)[source]¶ Plot precision recall curve for a specified brain.
- Parameters
data_dir (str) -- Path to directory where brain subvolumes are stored.
brain_id (str) -- Brain id to examine (brain2paths key from _data.py file).
object_type (str) -- soma or axon, the type of data to examine.
positive_channel (int) -- Channel that represents neuron in the predictions.
doubles (list, optional) -- Filenames of soma subvolumes that contain two somas, if applicable. Defaults to [].
show_plot (bool, optional) -- Whether to run pyplot, useful for pytests when figures should not be displayed. Defaults to True.
- Raises
ValueError -- _description_
- Returns
Best f-score across all thresholds. float: Threshold that yields the best validation f-score.
- Return type
-
brainlit.BrainLine.examine_threshold(data_dir: str, brain_id: str, threshold: float, object_type: str, positive_channel: int, dset: str = 'val', doubles: list = [], show_plot: bool = True)[source]¶ Display results in napari of all subvolumes that were below some performance threshold, at a given threshold.
- Parameters
data_dir (str) -- Path to directory where brain subvolumes are stored.
brain_id (str) -- Brain ID to examine (from _data.py file).
threshold (float) -- Threshold to examine.
object_type (str) -- soma or axon, the data type being examined.
positive_channel (int) -- 0 or 1, Channel that represents neuron in the predictions.
doubles (list, optional) -- Filenames of soma subvolumes that contain two somas, if applicable. Defaults to [].
show_plot (bool, optional) -- Whether to run napari, useful for pytests when figures should not be displayed. Defaults to True.
- Raises
ValueError -- If object_type is neither axon nor soma
ValueError -- If positive_channel is not 0 or 1.
-
class
brainlit.BrainLine.ApplyIlastik_LargeImage(ilastik_path: str, ilastik_project: str, ncpu: int, data_file: str, results_dir: Union[str, Path] = None)[source]¶ Apply ilastik to large image, where chunking is necessary.
- Parameters
ilastk_path (str) -- Path to ilastik executable.
ilastik_project (str) -- Path to ilastik project.
ncpu (int) -- Number of cpus to use for applying ilastik in parallel.
object_type (str) -- Soma for soma detection or axon for axon segmentaiton.
results_dir -- (str or Path): For soma detection, the directory to write detection results.
-
results_dir¶ (Path): For soma detection, the directory to write detection results.
-
apply_ilastik_parallel(self, brain_id: str, layer_names: list, threshold: float, data_dir: str, chunk_size: list, min_coords: list = [- 1, - 1, - 1], max_coords: list = [- 1, - 1, - 1])[source]¶ Apply ilastik to large brain, in parallel.
- Parameters
brain_id (str) -- Brain ID (key in brain2paths in _data.py file).
layer_names (list) -- Precomputed layer names to be appended to the base path.
threshold (float) -- Threshold for the ilastik predictor.
data_dir (str or Path) -- Path to directory where downloaded data will be temporarily stored.
chunk_size (list) -- Size of chunks to be used for parallel application of ilastik.
min_coords (list, optional) -- Lower bound of bounding box on which to apply ilastk (i.e. does not apply ilastik before these bounds). Defaults to [-1, -1, -1].
max_coords (list, optional) -- Upper bound of bounding box on which to apply ilastk (i.e. does not apply ilastik beyond these bounds). Defaults to [-1, -1, -1].
Results Analysis¶
-
class
brainlit.BrainLine.SomaDistribution(brain_ids: list, data_file: str, ontology_file: str, fixes_file: str = None, show_plots: bool = True, bootstrap: int = 0, shuffle: bool = False)[source]¶ Object to generate various analysis images for results from a set of brain IDs. An implementation of BrainDistribution class.
- Parameters
-
region_graph¶ Graph of region hierarchy with soma detection counts as node attributes.
- Type
nx.DiGraph
-
brainrender_somas(self, subtype_colors: dict, brain_region: str = 'DR')[source]¶ Generate brainrender viewer with soma detections.
- Parameters
subtype_colors (_type_) -- Mapping of subtypes (in soma_data.py file) to colors for soma plotting.
brain_region (str, optional) -- Brain region to display with the detections (e.g. dorsal raphe nucleus). Defaults to "DR".
-
napari_coronal_section(self, z: int, subtype_colors: dict, symbols: list = ['o', '+', '^', 'vbar'], fold_on: bool = False, plot_type: str = 'napari')[source]¶ Generate napari view with allen atlas and points of soma detections.
- Parameters
z (int) -- index of coronal slice in Allen atlas.
subtype_colors (dict) -- Mapping of subtypes (in soma_data.py file) to colors for soma plotting.
symbols (list) -- Napari point symbols to use for different samples of the same subtype. Defaults to ["o", "+", "^", "vbar"].
fold_on (bool, optional) -- Whether views should be a hemisphere, in which case detections from the other side are mirrored. Defaults to False.
plot_type (str, optional) -- Whether to plot via napari ('napari') or pyplot ('plt').
- Returns
pyplot figure only if plot_type=='plt'.
- Return type
pyplot.figure.Figure
-
region_barchart(self, regions: list, composite_regions: dict = {}, normalize_region: int = - 1)[source]¶ Generate bar charts comparing soma detection counts between regions.
- Parameters
regions (list) -- List of Allen atlas brain region IDs to display data for (ID's found here: http://api.brain-map.org/api/v2/structure_graph_download/1.json)
composite_regions (dict, optional) -- Mapping from a custom composite region (str, e.g. "Amygdala") to a set of regions that compose it (list of ints e.g. [131, 295, 319, 780]). Defaults to {}.
normalize_region (int, optional) -- Region ID to normalize data for the normalized bar chart. Defaults to -1.
-
class
brainlit.BrainLine.AxonDistribution(brain_ids: list, data_file: str, regional_distribution_dir: str, ontology_file: str, fixes_file: str = None, show_plots: bool = True)[source]¶ Generates visualizations of results of axon segmentations of a set of brains. Implements BrainDistribution.
- Parameters
brain_ids (list) -- List of brain IDs (keys of data json file).
data_files (str) -- Path to json file that contains information about samples.
regional_distribution_dir (str) -- Path to directory with pkl files that countain segmentation results by region.
show_plots (bool) -- Whether to show plots, only works if you have graphics access.
-
regional_distribution_dir¶ Path to directory with pkl files that countain segmentation results by region.
- Type
-
region_graph¶ Graph of region hierarchy with segmentation results as node attributes.
- Type
nx.DiGraph
-
brainrender_axons(self, subtype_colors: dict, brain_region: str = 'DR')[source]¶ Generate brainrender view to show axon segmentations.
- Parameters
subtype_colors (dict) -- Mapping of subtype to color to be used in visualization.
-
napari_coronal_section(self, z: int, subtype_colors: dict, fold_on: bool = False)[source]¶ Generate napari viewer with allen parcellation and heat map of axon segmentations.
-
region_barchart(self, regions: list, composite_regions: dict = {}, normalize_region: int = - 1)[source]¶ Generate bar charts with statistical tests to compare segmentations between brains.
- Parameters
regions (list) -- List of Allen atlas brain region IDs to display data for (ID's found here: http://api.brain-map.org/api/v2/structure_graph_download/1.json)
composite_regions (dict, optional) -- Mapping from a custom composite region (str, e.g. "Amygdala") to a set of regions that compose it (list of ints e.g. [131, 295, 319, 780]). Defaults to {}.
normalize_region (int, optional) -- Region ID to normalize data for the normalized bar chart. Defaults to -1.