library size normalization of inferred counts?

[vravik]

Hi Tinyi,

Quick question. Are the inferred cell-type specific expression matrices normalized for varying library size of samples/spots? In the vignette for downstream analysis you suggest performing a vst transformation, but this wouldn't correct for library size differences. I was just wondering if I should perform a counts normalization step of the inferred tumor-specific expression matrix, before using it for downstream analysis.

Just to clarify, my input bulk (spot-level) counts matrix is raw counts, as is the single cell reference.

Thank you, Best, Vish

[tinyi]

Hi Vish,

I believe vst transformation internally accounts for library size. You may see this for details: https://rdrr.io/bioc/DESeq2/man/varianceStabilizingTransformation.html . But you may certainly use log2 (library size normalized count + pseudo count) too.

Best,

Tinyi

On Mon, Feb 5, 2024 at 10:01 PM vravik @.***> wrote:

Hi Tinyi,

Quick question. Are the inferred cell-type specific expression matrices normalized for varying library size of samples/spots? In the vignette for downstream analysis you suggest performing a vst transformation, but this wouldn't correct for library size differences. I was just wondering if I should perform a counts normalization step of the inferred tumor-specific expression matrix, before using it for downstream analysis.

Just to clarify, my input bulk (spot-level) counts matrix is raw counts, as is the single cell reference.

Thank you, Best, Vish

— Reply to this email directly, view it on GitHub https://github.com/Danko-Lab/BayesPrism/issues/77, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB4NHS5N6KNDE5YSRLBQQSTYSGMGTAVCNFSM6AAAAABC3GYICGVHI2DSMVQWIX3LMV43ASLTON2WKOZSGEYTSOBUGE3TAMI . You are receiving this because you are subscribed to this thread.Message ID: @.***>