search

Danko-Lab / BayesPrism

A Fully Bayesian Inference of Tumor Microenvironment composition and gene expression
143 stars 45 forks source link

Doubts regarding cell state and cell label; log transformed input #15

Open Shubhra1994 opened 1 year ago

Shubhra1994 commented 1 year ago

Hi

Thanks for developing this. 1. I have a basic doubt regarding what cell state and cell label means. So, per my understanding cell label means what cell it is such as astrocyte, glia, etc What does the cell state mean here, should i cluster the single cell data such that the cells represent the state? Such that cells with similar transcriptional state represent similar state. Such that astrocyte and astrocyte like cell belong to state 1 and so on ?

To circumvent this I made the cell label as cell state and ran the program, is this okay?

  1. However when I run the program I get Warning: input seems to be log-transformed. Please double check your input. Log transformation should be avoided

However I am choosing the counts slot of seurat object which seems to contain raw counts. Is there any way I can check if this is the issue with the dataset or why exactly this error comes up?

Thanks again!

tinyi commented 1 year ago

Hi.

Thank you for your interest in our work.

To answer your question: cell type is your desired granularity, while cell state is the very fine granularity where there is a great extent of uncertainty that we want to marginalize (for example, to be used to model subclusters of cells lie on a continuum or have some degree of heterogeneity such as the cancer cells). The definition can be a bit problem-specific.

Please note that cell state is never a required input, if cell state is not provided, BayesPrism will treat cell state one-to-one correspond to cell types.

Hope that this helps.

Best,

Tinyi

On Mon, Sep 12, 2022 at 7:52 AM Shubhra1994 @.***> wrote:

Hi

Thanks for developing this. I have a basic doubt regarding what cell state and cell label means.

So, per my understanding cell label means what cell it is such as astrocyte, glia, etc What does the cell state mean here, should i cluster the single cell data such that the cells represent the state?

Since, the atlas is from human cell atlas any idea how I can indicate the cell state from this?

Thanks again!

— Reply to this email directly, view it on GitHub https://github.com/Danko-Lab/BayesPrism/issues/15, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB4NHS7SKIYLJHD5GD2IE3TV54KORANCNFSM6AAAAAAQKMOPUQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>