spanorm_pca

spanorm_pca(
    data,
    nsvgs=3000,
    ncomponents=50,
    svg_fdr=1.0,
    residuals='deviance',
    key_added='X_spanorm_pca',
    svd_solver='exact',
    n_jobs=1,
    random_state=0,
    layer=None,
    table_key=None,
    copy=False,
)

GLM-based (SpaNorm) PCA (R SpaNormPCA).

Deviance or Pearson residuals of the stored technical fit (spanorm_null, created by :func:~spanormpy.spanorm_svg) are computed for the top nsvgs SVGs (see :func:~spanormpy.top_svgs), centred per gene (not scaled), and decomposed.

Parameters

Name	Type	Description	Default
data		An AnnData or SpatialData on which spanorm and spanorm_svg have run.	required
nsvgs	int	Number of SVGs used, number of components, and the SVG FDR threshold.	3000
ncomponents	int	Number of SVGs used, number of components, and the SVG FDR threshold.	3000
svg_fdr	int	Number of SVGs used, number of components, and the SVG FDR threshold.	3000
residuals	str	"deviance" (default) or "pearson".	'deviance'
key_added	str	Scores go to obsm[key_added]; uns[key_added] holds variance (squared standard deviations), percent_var (R percentVar, percent of the variance of the computed components), svgs (the genes used) and the settings; varm[key_added + "_loadings"] holds the rotation (R rotation), zero for genes not used.	'X_spanorm_pca'
svd_solver	str	"exact" (numpy.linalg.svd, R ExactParam) or "randomized" (randomised SVD for large data; approximate; seeded by random_state).	'exact'
n_jobs	int	No effect; accepted for compatibility with R’s BPPARAM.	1
random_state	int	Seed of the randomised SVD (svd_solver="randomized" only).	0
layer	str | None	Counts layer (default X) and, for SpatialData, the table.	None
table_key	str | None	Counts layer (default X) and, for SpatialData, the table.	None
copy	bool	Return a modified copy instead of modifying data in place.	False

Returns

Name	Type	Description
	None, or with copy=True a modified copy of the same container type.

Examples

Note

The code below runs when the site is built, on a small pre-fitted subset of HumanDLPFC (301 genes x 4015 spots; spanorm(sample_p=0.05, df_tps=2), spanorm_svg() and spanorm_pca() already applied) that is fitted once per build and loaded as adata. The outputs are those of the code shown.

import spanormpy as sn
# adata: the HumanDLPFC AnnData (raw counts in X, coordinates in obsm["spatial"]),
# already fitted with sn.spanorm(adata, sample_p=0.05, df_tps=2)
sn.spanorm_pca(adata, ncomponents=10)
adata.obsm["X_spanorm_pca"].shape

(4015, 10)