spared.layer_operations.combat_transformation

spared.layer_operations.combat_transformation(adata: AnnData, batch_key: str, from_layer: str, to_layer: str) → AnnData[source]

Perform batch correction with ComBat

Compute batch correction using the pycombat package. The batches are defined by adata.obs[batch_key] so the user can define which variable to use as batch identifier. The input data for the batch correction is adata.layers[from_layer] and the output is stored in adata.layers[to_layer]. Importantly, as the original ComBat paper notes the data should be approximately normally distributed. Therefore, it is recommended to use this function over \(\log_2(TPM+1)\) data.

Parameters:

  • adata (ad.AnnData) – The AnnData object to transform. Must have logarithmically transformed data in adata.layers[from_layer].

  • batch_key (str) – The column in adata.obs that defines the batches.

  • from_layer (str) – The layer to take the data from.

  • to_layer (str) – The layer to store the results of the transformation.

Returns:

The updated AnnData object with batch corrected data in adata.layers[to_layer].

Return type:

ad.AnnData