Visium with SCT Transformation and Harmony Integration

[dimurali93]

Hi All,
Thanks in advance for the help. I have 5 visium samples that I am analyzing using SCT and Hamony integration. I have the following questions.

My pipeline :

#QC
merged[["percent.mt"]] <- PercentageFeatureSet(merged,   pattern = "^MT-*", assay = "Spatial")
merged[["percent_hb"]] <- PercentageFeatureSet(merged, pattern ="^HB.-*", assay = "Spatial")
merged <- merged[, merged$nFeature_Spatial > 500 &  merged$nCount_Spatial > 200 &
                         merged$percent.mt < 8 & merged$percent_hb < 10]
merged <- merged[!grepl("^ALB", rownames(merged)), ]
merged <- merged[!grepl("^HB.*-", rownames(merged)), ]

#SCT
merged_sct <- SCTransform(merged,
                             assay = "Spatial",
                             vst.flavor = "v2",
                             verbose = FALSE) %>%
  RunPCA(npcs = 30,  verbose = FALSE) %>%
  RunUMAP(reduction = "pca", dims = 1:30, verbose = FALSE) %>%
  FindNeighbors(reduction = "pca", dims = 1:30, verbose = FALSE) %>%
  FindClusters(resolution = 0.5,verbose = FALSE)

options(future.globals.maxSize = 3e+09)

#INTEGRATION
METHOD 1

integrated_SCT_harmony_1 <- IntegrateLayers(object = merged_sct, method = HarmonyIntegration,
orig.reduction = "pca", new.reduction = 'harmony',
assay = "SCT", verbose = FALSE)

pct <- integrated_SCT_harmony_1[["pca"]]@stdev / sum(integrated_SCT_harmony_1[["pca"]]@stdev) * 100
cumu <- cumsum(pct)
co1 <- which(cumu > 90 & pct < 5)[1]
co2 <- sort(which((pct[1:length(pct) - 1] - pct[2:length(pct)]) > 0.1), decreasing = T)[1] + 1
pcs <- min(co1, co2)
integrated_SCT_harmony_1 <- FindNeighbors(integrated_SCT_harmony_1, dims = 1:pcs, reduction = "harmony")
integrated_SCT_harmony_1 <- FindClusters(integrated_SCT_harmony_1, resolution = 0.5)
integrated_SCT_harmony_1 <- RunUMAP(integrated_SCT_harmony_1, dims = 1:pcs, reduction = "harmony")

integrated_SCT_harmony_1[["Spatial"]] <- JoinLayers(integrated_SCT_harmony_1[["Spatial"]])
integrated_SCT_harmony_1 = PrepSCTFindMarkers(integrated_SCT_harmony_1, assay = "SCT", verbose = TRUE)

> Found 5 SCT models. Recorrecting SCT counts using minimum median counts: 5454.5
> Error in PrepSCTFindMarkers(integrated_SCT_harmony_2, assay = "SCT", verbose = TRUE) : 
>   Multiple UMI assays are used for SCTransform: Spatial, RNA

METHOD2

integrated_SCT_harmony_2<- IntegrateLayers( object = merged_sct, method = HarmonyIntegration,
 new.reduction = "harmony", normalization.method = "SCT", verbose = TRUE)
pct <- integrated_SCT_harmony_2[["pca"]]@stdev / sum(integrated_SCT_harmony_2[["pca"]]@stdev) * 100
 cumu <- cumsum(pct)
 co1 <- which(cumu > 90 & pct < 5)[1]
 co2 <- sort(which((pct[1:length(pct) - 1] - pct[2:length(pct)]) > 0.1), decreasing = T)[1] + 1
 pcs <- min(co1, co2)
integrated_SCT_harmony_2 <- FindNeighbors(integrated_SCT_harmony_2, dims = 1:pcs, reduction = "harmony")
integrated_SCT_harmony_2 <- FindClusters(integrated_SCT_harmony_2, resolution = 0.5)
integrated_SCT_harmony_2 <- RunUMAP(integrated_SCT_harmony_2, dims = 1:pcs, reduction = "harmony")

integrated_SCT_harmony_2[["Spatial"]] <- JoinLayers(integrated_SCT_harmony_2[["Spatial"]])
integrated_SCT_harmony_2 = PrepSCTFindMarkers(integrated_SCT_harmony_2, assay = "SCT", verbose = TRUE)

> Found 5 SCT models. Recorrecting SCT counts using minimum median counts: 5454.5
> Error in PrepSCTFindMarkers(integrated_SCT_harmony_2, assay = "SCT", verbose = TRUE) : 
>   Multiple UMI assays are used for SCTransform: Spatial, RNA

1. What is the correct syntax for integration of SCT object using harmony? ( code taken from [Integrating datasets with SCTransform in Seurat v5 #7542] and documentation )(Integrating datasets with SCTransform in Seurat v5 #7542))

integrated_SCT_harmony_1<- IntegrateLayers(object = merged_sct, method = HarmonyIntegration, orig.reduction = "pca", new.reduction = 'harmony', assay = "SCT", verbose = TRUE)

or

integrated_SCT_harmony_2<- IntegrateLayers( object = merged_sct, method = HarmonyIntegration, new.reduction = "harmony", normalization.method = "SCT", verbose = TRUE)

2.Do I have to still run JoinLayers() and PrepSCTFindMarkers() after Harmony integration to do downstream analysis based on DE genes?
If so I am getting the following error:

integrated_SCT_harmony <- JoinLayers(integrated_SCT_harmony, assay = "Spatial")
integrated_SCT_harmony = PrepSCTFindMarkers(integrated_SCT_harmony, assay = "SCT", verbose = TRUE)

Found 5 SCT models. Recorrecting SCT counts using minimum median counts: 5454.5
Error in PrepSCTFindMarkers(integrated_SCT_harmony_2, assay = "SCT", verbose = TRUE) :
Multiple UMI assays are used for SCTransform: Spatial, RNA

Can I run FindMarkers without PrepSCTFindMarerks in seurat V5? Could you also please shed light into when to use SCT assay versus when to use Spatial assay here? for instance should i join the spatial or the SCT assay layers? should i run the PrepSCTFindMarkers for SCT assay?

3. What should the default assay be after SCT transformation and before harmony integration?

Thanks,
Divya Murali

[alifarhat40]

Hi. Did you ever figure this out?