search

juexinwang / scGNN

scGNN (single cell graph neural networks) for single cell clustering and imputation using graph neural networks
MIT License
133 stars 40 forks source link

running scGNN on integrated datasets #29

Open iichelhadi opened 7 months ago

iichelhadi commented 7 months ago

Hello,

I am very interested in using your tool for my analyses but I was wondering if it can be used on integrated data. I am performing analyses on scRNA from different studies and sequencing techniques which have different sequencing depths, unfortunately. If not on the gene expression matrix could I do it on the batch-corrected latent representation?

Regards

juexinwang commented 7 months ago

Batch effect correction method is highly recommended before you input the data to scGNN.

iichelhadi commented 7 months ago

Batch effect correction method is highly recommended before you input the data to scGNN.

Thanks but this doesn't really answer my question. Should I used normalized and scaled counts that are batch corrected or dim reduction cell.embeddings which are also batch corrected?

Regards

juexinwang commented 7 months ago

Not exactly, batch effect is annoying and needs extra effort with other methods. See this paper: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1850-9

On Tue, Feb 27, 2024 at 5:29 AM iichelhadi @.***> wrote:

Batch effect correction method is highly recommended before you input the data to scGNN.

Thanks but this doesn't really answer my question. Should I used normalized and scaled counts that are batch corrected or dim reduction cell.embeddings which are also batch corrected?

Regards

— Reply to this email directly, view it on GitHub https://github.com/juexinwang/scGNN/issues/29#issuecomment-1965632724, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAZPXTAOALVV4F5AVCSXGVDYVUZ5BAVCNFSM6AAAAABDZRIKTWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNRVGYZTENZSGQ . You are receiving this because you commented.Message ID: @.***>