`spateo.alignment.methods.morpho`#

Module Contents#

Functions#

`con_K`(→ Union[numpy.ndarray, torch.Tensor])	con_K constructs the Squared Exponential (SE) kernel, where K(i,j)=k(X_i,Y_j)=exp(-beta*\|\|X_i-Y_j\|\|^2).
`get_P`(→ Tuple[Any, Any, Any])	Calculating the generating probability matrix P.
`get_P_chunk`(→ Union[numpy.ndarray, torch.Tensor])	Calculating the generating probability matrix P.
`BA_align`(→ Tuple[Optional[Tuple[anndata.AnnData, ...)	_summary_

spateo.alignment.methods.morpho.con_K(X: numpy.ndarray | torch.Tensor, Y: numpy.ndarray | torch.Tensor, beta: int | float = 0.01, use_chunk: bool = False) → numpy.ndarray | torch.Tensor[source]#

con_K constructs the Squared Exponential (SE) kernel, where K(i,j)=k(X_i,Y_j)=exp(-beta*||X_i-Y_j||^2).

Parameters:

X: The first vector Xinmathbb{R}^{N imes d}
Y: The second vector Xinmathbb{R}^{M imes d}
beta: The length-scale of the SE kernel.
use_chunk bool, optional: Whether to use chunk to reduce the GPU memory usage. Note that if set to ``True’’ it will slow down the calculation. Defaults to False.

Returns:

The kernel Kinmathbb{R}^{N imes M}

Return type:

Calculating the generating probability matrix P.

Parameters:

XAHat: Current spatial coordinate of sample A. Shape: N x D.
XnB: spatial coordinate of sample B (reference sample). Shape: M x D.
sigma2: The spatial coordinate noise.
beta2: The gene expression noise.
alpha: A vector that encoding each probability generated by the spots of sample A. Shape: N x 1.
gamma: Inlier proportion of sample A.
Sigma: The posterior covariance matrix of Gaussian process. Shape: N x N or N x 1.
GeneDistMat: The gene expression distance matrix between sample A and sample B. Shape: N x M.
SpatialDistMat: The spatial coordinate distance matrix between sample A and sample B. Shape: N x M.
samples_s: The space size of each sample. Area size for 2D samples and volume size for 3D samples.

Returns:

Generating probability matrix P. Shape: N x M.

Return type:

Calculating the generating probability matrix P.

Parameters:

XAHat: Current spatial coordinate of sample A. Shape

spateo.alignment.methods.morpho.BA_align(sampleA: anndata.AnnData, sampleB: anndata.AnnData, genes: List | torch.Tensor | None = None, spatial_key: str = 'spatial', key_added: str = 'align_spatial', iter_key_added: str | None = 'iter_spatial', vecfld_key_added: str | None = 'VecFld_morpho', layer: str = 'X', dissimilarity: str = 'kl', keep_size: bool = False, max_iter: int = 200, lambdaVF: int | float = 100.0, beta: int | float = 0.01, K: int | float = 15, normalize_c: bool = True, normalize_g: bool = True, select_high_exp_genes: bool | float | int = False, dtype: str = 'float32', device: str = 'cpu', inplace: bool = True, verbose: bool = True, nn_init: bool = True, SVI_mode: bool = True, batch_size: int = 1000, partial_robust_level: float = 25) → Tuple[Tuple[anndata.AnnData, anndata.AnnData] | None, numpy.ndarray, numpy.ndarray][source]#

_summary_

Parameters:

sampleA: Sample A that acts as reference.
sampleB: Sample B that performs alignment.
genes: Genes used for calculation. If None, use all common genes for calculation.
spatial_key: The key in .obsm that corresponds to the raw spatial coordinate.
key_added: .obsm key under which to add the aligned spatial coordinate.
iter_key_added: .uns key under which to add the result of each iteration of the iterative process. If iter_key_added is None, the results are not saved.
vecfld_key_added: The key that will be used for the vector field key in .uns. If vecfld_key_added is None, the results are not saved.
layer: If 'X', uses .X to calculate dissimilarity between spots, otherwise uses the representation given by .layers[layer].
dissimilarity: Expression dissimilarity measure: 'kl' or 'euclidean'.
small_variance: When approximating the assignment matrix, if True, we use small sigma2 (0.001) rather than the infered sigma2
max_iter: Max number of iterations for morpho alignment.
lambdaVF: Hyperparameter that controls the non-rigid distortion degree. Smaller means more flexibility.
beta: The length-scale of the SE kernel. Higher means more flexibility.
K: The number of sparse inducing points used for Nystr ̈om approximation. Smaller means faster but less accurate.
normalize_c: Whether to normalize spatial coordinates.
normalize_g: Whether to normalize gene expression. If dissimilarity == 'kl', normalize_g must be False.
select_high_exp_genes: Whether to select genes with high differences in gene expression.
samples_s: The space size of each sample. Area size for 2D samples and volume size for 3D samples.
dtype: The floating-point number type. Only float32 and float64.
device: Equipment used to run the program. You can also set the specified GPU for running. E.g.: '0'.
inplace: Whether to copy adata or modify it inplace.
verbose: If True, print progress updates.
nn_init: If True, use nearest neighbor matching to initialize the alignment.
SVI_mode: Whether to use stochastic variational inferential (SVI) optimization strategy.
batch_size: The size of the mini-batch of SVI. If set smaller, the calculation will be faster, but it will affect the accuracy, and vice versa. If not set, it is automatically set to one-tenth of the data size.
partial_robust_level: The robust level of partial alignment. The larger the value, the more robust the alignment to partial cases is. Recommended setting from 1 to 50.

spateo.alignment.methods.morpho#

Module Contents#

Functions#

`spateo.alignment.methods.morpho`#