Suggestion to specify library.size parameter in NormalizeData()

[shiquan]

Dear Developers,

Thank you for creating and maintaining this excellent package. I greatly appreciate the effort that has gone into developing Seurat, as it has been invaluable for my research. I have also developed methods based on Seurat, and I would like to share a small suggestion regarding the NormalizeData() function.

Seurat's flexibility in accepting different assays, such as exon, junction, and other types alongside gene expression data is a powerful feature. However, I noticed a potential issue when normalizing data across these assays. The default normalization method adjusts values using the library size, which is typically calculated as the sum of feature counts for each cell. While this works well for single assays, the library size can vary for the same cell across different assays, making it challenging to compare expression levels between assays. For example, for the feature plot of an exon and its gene, the data from layer 'data', the exon expression might exceed gene expression level for the same cell.

To address this, I suggest adding a new library.size parameter to the NormalizeData() function. This would allow users to specify a consistent library size, such as the gene count (nCount_RNA by default) from the metadata table, for normalization across all assays.

I believe this enhancement would make Seurat even more robust for multi-assay analysis.

Thank you for considering this suggestion, and please let me know if further clarification or discussion would be helpful.

Bests,
Quan Shi