arcfish.pp.filter_normalize

arcfish.pp.filter_normalize(adata: AnnData)

Filter outliers and normalized by 1D genomic distance.

Filter out entries with pairwise difference 4 standard deviations away from the median pairwise difference stratified by 1D genomic distance. The standard deviations are estimated from a local linear regression.

The filtered pairwise difference is then normalized by the standard deviations stratified by 1D genomic distance. Similarly, the standard deviations are estimated from a local linear regression.

Append the followings to the varp field of adata:

1. raw_var_{X,Y,Z}: median squared pairwise difference of each axis.

2. var_{X,Y,Z}: mean squared pairwise difference of each axis after normalization.

3. count_{X,Y,Z}: the number of available values for each pairwise distance entry after normalization.

Parameters:

adata (AnnData) – Object created by FOF_CT_Loader.create_adata().