Run harmony by calling IntegrateLayers - how to specify the batch covariate in meta.data?

[linpei26]

Hello!

I find the tutorial at https://satijalab.org/seurat/articles/seurat5_integration and learn that we could call harmony by the function "IntegrateLayers". But in the example code (copied here below), I did not see any parameter to specify batch covariate.

obj <- IntegrateLayers(
  object = obj, method = HarmonyIntegration,
  orig.reduction = "pca", new.reduction = "harmony",
  verbose = FALSE
)

So, is it unnecessary to use such a para when applying IntegrateLayers, or am I missing anything? Thanks in advance. Best, P

[adairama]

The first half of the vignette talks about the data processing steps. The very first step is to split the unintegrated dataset by the batch covariate (it is "Method" in the vignette). Here is a minimal example inspired by the vignette:

## Step 1: Split by the batch covariate
obj[["RNA"]] <- split(obj[["RNA"]], f = obj$Method)

## Step 2: Preprocess each split
obj <- obj %>% 
   NormalizeData(verbose = FALSE) %>%
   FindVariableFeatures(verbose = FALSE)  %>%
   ScaleData(verbose = FALSE) %>%
   RunPCA(verbose = FALSE)

## Step 3: Data integration and generate UMAP
obj <- IntegrateLayers(obj, 
   method = HarmonyIntegration,
   orig.reduction = "pca", 
   new.reduction = "harmony",
   verbose = FALSE)

obj <- RunUMAP(obj, reduction = "harmony", dims = 1:30, reduction.name = "umap.harmony")

## Step 4: Optional cleanup
obj <- JoinLayers(obj)
obj[["RNA"]]$scale.data <- NULL

[linpei26]

Thank you. I am now using the harmony package directly.

The code is:

obj_1 = merge(x=o1,y=o2)

obj_1 = NormalizeData(obj_1)
obj_1 = FindVariableFeatures(obj_1)
obj_1 = ScaleData(obj_1)

obj_1 = RunPCA(obj_1)

obj_1 = RunHarmony(obj_1, c("dataset","flowcell"))
obj_1[["RNA"]] = JoinLayers(obj_1[["RNA"]])

obj_1 = FindNeighbors(obj_1, dims = 1:20, reduction = "harmony")
obj_1 = FindClusters(obj_1)
obj_1 = RunUMAP(obj_1, dims = 1:20, reduction = "harmony")

Please feel free to me know if you have any suggestions. Thanks. P