CAJAL: a Python package for the analysis of single-cell morphological data
CAJAL is a Python package designed to explore and analyze the morphology of cells and its relationship with other single-cell data.
CAJAL leverages the Python Optimal Transport library to compute the Gromov-Wasserstein (GW) distance between every pair of cells in a given sample. This distance quantifies the degree to which the shape of one cell can be transformed into that of another with minimal stretching or bending. One of the key benefits of using the GW distance is that it does not require any prior knowledge or model for the morphology of the cells. This feature makes CAJAL suitable for studying arbitrarily heterogeneous mixtures of cells with highly complex and diverse morphologies that may defy straightforward classification.
The morphological distance produced by CAJAL is a bona-fide mathematical distance in a latent space of cell morphologies. In this latent space, each cell is represented by a point, and distances between cells indicate the amount of physical deformation needed to change the morphology of one cell into that of another. By formulating the problem in this way, CAJAL can make use of standard statistical and machine learning approaches to define cell populations based on their morphology; dimensionally reduce and visualize cell morphology spaces; and integrate cell morphology spaces across tissues, technologies, and with other single-cell data modalities, among other analyses.
- What is CAJAL?
- Computing Intracellular Distance Matrices
- Computing GW Distances
- Inferring Associations with Cell Morphology
- Computing Average Cell Shapes
- Tutorial 1: Predicting the Molecular Type of Neurons
- Tutorial 2: Genetic Determinants of Neuronal Morphology
- Tutorial 3: Computing Morphological Distances in Large Datasets
- Processing SWC Files
NeuronNode
NeuronTree
cajal.swc.SWCForest
read_swc()
linearize()
forest_from_linear()
write_swc()
default_name_validate()
cell_iterator()
keep_only_eu()
preprocessor_geo()
preprocessor_eu()
total_length()
weighted_depth()
discrete_depth()
node_type_counts_tree()
node_type_counts_forest()
num_nodes()
read_preprocess_save()
get_filenames()
batch_filter_and_preprocess()
WeightedTreeRoot
WeightedTreeChild
WeightedTree_of()
- Sampling from SWC Files
- Processing Obj Meshes
- Sampling from Segmented Images
- Running Gromov-Wasserstein
- Second Lower Bound and Quantized Gromov-Wasserstein
- Laplacian Score
- Average Cell Shapes
- Clustering