This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Seurat V5 integration

Hello. I have two seurat objects. One is the infected sample and the second is the uninfected. Here is the code:

merged_dat <- list(inf,uninf)
merged_dat[[1]] <- subset(merged_dat[[1]], subset = nFeature_RNA > 200 & nFeature_RNA < 8200 & percent.mt < 10)
merged_dat[[2]] <- subset(merged_dat[[2]], subset = nFeature_RNA > 200 & nFeature_RNA < 10000 & percent.mt < 10)

for (x in seq_along(merged_dat)) {  
  merged_dat[[x]] <- SCTransform(merged_dat[[x]],verbose = FALSE)
}

So I want to integrate them using Seurat V5 as explained in this link: https://satijalab.org/seurat/articles/seurat5_integration

But how do I combine the two seurat objects into one to run the integration code such as this one:

obj <- IntegrateLayers(
  object = obj, method = CCAIntegration,
  orig.reduction = "pca", new.reduction = "integrated.cca", verbose = FALSE
)
seurat integration
• 3,759 views • link
updated by Ram Ram • written by starswillfade • starswillfade

1 answer

It will depend on your context and question.

I guess you will want to compare gene expression between your conditions within cell types at some point. If you integrate your samples this covariate will be reduced.

Just merge the sample and look at your cell projection on UMAP or TSNE to see if it make sense in your biological question. For example if you expect a certain cell type to not be affected by your infection, cells from this cell type should cluster together (infected and uninfected cells) on the projection.

merged_dat <- list(inf,uninf)
merged_dat[[1]] <- subset(merged_dat[[1]], subset = nFeature_RNA > 200 & nFeature_RNA < 8200 & percent.mt < 10)
merged_dat[[2]] <- subset(merged_dat[[2]], subset = nFeature_RNA > 200 & nFeature_RNA < 10000 & percent.mt < 10)
data <- merge(merged_dat[[1]], y = merged_dat[[2]], add.cell.ids = c("inf", "uninf"), project = "name", merge.data = TRUE)

You can then scale the new data object via ScaleData or SCTransform and continue your analysis by reducing dimensionality and clustering.

• 0 views • link
written by Bastien Hervé Bastien Hervé

Log in to answer this question.

More posts like this