Log Transformation Neccessity of SCnormed data

[YSPeng1225]

Hi,

Thanks for providing such an effective scRNA normalization method, and we are applying it to our newest research!

However, when applying different normalization methods, I noticed that other approaches, such as TC, which equals t(t(counts)/colSums(counts) * mean(colSums(counts))), will be followed by a log transformation to alleviate the effects of extreme values, such as log2(.+1)/log1p().

My confusion is if I normalize the counts data frame through SCnorm, would it be necessary to log transform it as well?

Here is my code for your reference:

sce <- SCnorm(Data = raw, # this is the counts matrix
Conditions = groups,
PrintProgressPlots = FALSE,
FilterCellNum = 10,
NCores=ncore, reportSF = FALSE,
ditherCounts=TRUE)
raw_normed <- SingleCellExperiment::normcounts(sce)

raw_normed <- log2(raw_normed+1)

I would appreciate it if you could help me resolve it.

Thanks,
Yuansheng

[rhondabacher]

Dear Yuansheng, Thank you! This depends on what you're planning to do downstream of the normalization. If you're planning to do a dimension reduction and visualize then yes, that is generally advised. If you are planning to do DE analysis, again this will depend on what method you're planning to use. If a method expects raw counts, you can output the SCnorm normalization factors rather than normalizing the data directly by setting reportSF=TRUE Best, Rhonda

…

On Mon, Apr 28, 2025 at 5:49 AM YSPeng ***@***.***> wrote: *YSPeng1225* created an issue (rhondabacher/SCnorm#39) <#39> Hi, Thanks for providing such an effective scRNA normalization method, and we are applying it to our newest research! However, when applying different normalization methods, I noticed that other approaches, such as TC, which equals t(t(counts)/colSums(counts) * mean(colSums(counts))), will be followed by a log transformation to alleviate the effects of extreme values, such as log2(.+1)/log1p(). My confusion is if I normalize the counts data frame through SCnorm, would it be necessary to log transform it as well? Here is my code for your reference: ` sce <- SCnorm(Data = raw, # this is the counts matrix Conditions = groups, PrintProgressPlots = FALSE, FilterCellNum = 10, NCores=ncore, reportSF = FALSE, ditherCounts=TRUE) raw_normed <- SingleCellExperiment::normcounts(sce) raw_normed <- log2(raw_normed+1) ` I would appreciate it if you could help me resolve it. Thanks, Yuansheng — Reply to this email directly, view it on GitHub <#39>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABSOD3EXYYX5EZAC4BTQT7323YBRZAVCNFSM6AAAAAB4ACQ6AWVHI2DSMVQWIX3LMV43ASLTON2WKOZTGAZDINJTGI4DKMY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Rhonda Bacher, PhD Associate Professor Department of Biostatistics University of Florida Office: 460 Dauer Hall Phone: 352-294-5914 Email: ***@***.*** http://rhondabacher.com/