INTEGRATION_CAMPARISION

1 Functions

2 Data

3 13 clusters

GBMSeurat_cancer_sipsic <- FindClusters(GBMSeurat_cancer_sipsic, resolution = 0.9)
GBMSeurat_cancer_sipsic_cp <- FindClusters(GBMSeurat_cancer_sipsic_cp, resolution = 0.65)
GBM.combined <- FindClusters(GBM.combined, resolution = 0.8)
GBMSeurat_cancer_harmony <- FindClusters(GBMSeurat_cancer_harmony, resolution = 0.5)
GBMSeurat_cancer  %<>% FindClusters(resolution  = 0.2)

GBMSeurat_cancer@project.name = "GBM_original"
GBMSeurat_cancer_sipsic_cp@project.name = "GBM_sipsic_cp"
GBM.combined@project.name = "GBM_seurat_integration"
GBMSeurat_cancer_harmony@project.name = "GBM_harmony"

all_datasets = list(GBMSeurat_cancer,GBMSeurat_cancer_sipsic,GBMSeurat_cancer_sipsic_cp,GBM.combined,GBMSeurat_cancer_harmony)

3.1 UMAPS

for (dataset in all_datasets) {
  print_tab(DimPlot(dataset, reduction = "umap",group.by = c("seurat_clusters","orig.ident","cancer_type"),ncol = 2),title = dataset@project.name,subtitle_num = 3)
}

GBM_original

GBM_sipsic

GBM_sipsic_cp

GBM_seurat_integration

GBM_harmony

NA

3.2 stacked batplot

for (dataset in all_datasets) {
  
  clusters_and_scores = FetchData(object = dataset,vars= c("cancer_type","seurat_clusters")) %>%  group_by(seurat_clusters,cancer_type) %>%
    summarise(n_cells = n(), .groups = "drop_last")%>% mutate(per =  100 *n_cells/sum(n_cells))
  
  integration_score = clusters_and_scores %>%  group_by(seurat_clusters) %>% filter(n_cells == max(n_cells)) %>% pull(per) %>% mean() %>% round(digits = 2)
  integration_score_sum_cells = clusters_and_scores %>%  group_by(seurat_clusters) %>% filter(n_cells == max(n_cells)) %>% pull(n_cells) %>% sum() %>%
    divide_by(sum(clusters_and_scores$n_cells)) %>% round(digits = 2)
  
  v_factor_levels <-c( "MesLike1", "MesLike2", "NPCLike1", "NPCLike2", "OPCLike","ACLike")
  colors = RColorBrewer::brewer.pal(6, "Paired"); colors[5] = "orange"
  p2 = ggplot(data=clusters_and_scores, aes(x=seurat_clusters, y=per, fill=factor(cancer_type, levels = v_factor_levels))) +
    geom_bar(stat="identity")+theme_minimal() + scale_fill_manual(values = colors,name  = "Cancer type")+ 
    labs(title = dataset@project.name,subtitle = "integration score=" %s+% integration_score %s+% "%" %s+% "\nintegration sums cells score="
         %s+% integration_score_sum_cells)+
    ylab("% from cluster")
  
  
  clusters_and_scores = FetchData(object = dataset,vars= c("orig.ident","seurat_clusters")) %>%  group_by(seurat_clusters,orig.ident) %>%  
    summarise(n_cells = n(), .groups = "drop_last")%>% mutate(per =  100 *n_cells/sum(n_cells))
  
  colors = RColorBrewer::brewer.pal(9, "Paired")
  p3 = ggplot(data=clusters_and_scores, aes(x=seurat_clusters, y=per, fill=factor(orig.ident))) +
    geom_bar(stat="identity")+theme_minimal() + scale_fill_manual(values = colors,name  = "Patient")+ 
    labs(title = dataset@project.name)+
    ylab("% from cluster")
  
  print_tab(p2+p3,title = dataset@project.name,subtitle_num = 3)
}

GBM_original

GBM_sipsic

GBM_sipsic_cp

GBM_seurat_integration

GBM_harmony

NA

3.3 Combine cancer subtypes

for (i in seq_along(all_datasets)) {
  all_datasets[[i]]$cancer_type_combined = all_datasets[[i]]$"cancer_type"  %>% gsub(pattern = "MesLike1|MesLike2",replacement = "MesLike")%>% gsub(pattern = "NPCLike1|NPCLike2",replacement = "NPCLike")
}

for (dataset in all_datasets) {
  clusters_and_scores = FetchData(object = dataset,vars= c("cancer_type_combined","seurat_clusters")) %>%  group_by(seurat_clusters,cancer_type_combined) %>%  summarise(n_cells = n(), .groups = "drop_last")%>% mutate(per =  100 *n_cells/sum(n_cells))
  
  integration_score = clusters_and_scores %>%  group_by(seurat_clusters) %>% filter(n_cells == max(n_cells)) %>% pull(per) %>% mean() %>% round(digits = 2)
  
  integration_score_sum_cells = clusters_and_scores %>%  group_by(seurat_clusters) %>% filter(n_cells == max(n_cells)) %>% pull(n_cells) %>% sum() %>% 
    divide_by(sum(clusters_and_scores$n_cells)) %>% round(digits = 2)
  
  v_factor_levels <-c( "MesLike", "NPCLike", "OPCLike","ACLike")
  colors = RColorBrewer::brewer.pal(6, "Paired")[c(2,4,5,6)]; colors[3] = "orange"
  p4 = ggplot(data=clusters_and_scores, aes(x=seurat_clusters, y=per, fill=factor(cancer_type_combined, levels = v_factor_levels))) +
    geom_bar(stat="identity")+theme_minimal() + scale_fill_manual(values = colors,name  = "Cancer type")+ 
    labs(title = dataset@project.name,subtitle = "integration score=" %s+% integration_score %s+% "%" %s+% "\nintegration sums cells score=" %s+% 
           integration_score_sum_cells)+ylab("% from cluster")
  print_tab(p4,title = dataset@project.name,subtitle_num = 3)
}

GBM_original

GBM_sipsic

GBM_sipsic_cp

GBM_seurat_integration

GBM_harmony

NA

4 10 clusters

GBMSeurat_cancer_sipsic <- FindClusters(GBMSeurat_cancer_sipsic, resolution = 0.65)
GBMSeurat_cancer_sipsic_cp <- FindClusters(GBMSeurat_cancer_sipsic_cp, resolution = 0.4)
GBM.combined <- FindClusters(GBM.combined, resolution = 0.6)
GBMSeurat_cancer_harmony <- FindClusters(GBMSeurat_cancer_harmony, resolution = 0.19)
GBMSeurat_cancer  %<>% FindClusters(resolution  = 0.09)

all_datasets = list(GBMSeurat_cancer,GBMSeurat_cancer_sipsic,GBMSeurat_cancer_sipsic_cp,GBM.combined,GBMSeurat_cancer_harmony)

4.1 UMAPS

for (dataset in all_datasets) {
  print_tab(DimPlot(dataset, reduction = "umap",group.by = c("seurat_clusters","orig.ident","cancer_type"),ncol = 2),title = dataset@project.name,subtitle_num = 3)
}

GBM_original

GBM_sipsic

GBM_sipsic_cp

GBM_seurat_integration

GBM_harmony

NA

4.2 stacked batplot

for (dataset in all_datasets) {
  
  clusters_and_scores = FetchData(object = dataset,vars= c("cancer_type","seurat_clusters")) %>%  group_by(seurat_clusters,cancer_type) %>%
    summarise(n_cells = n(), .groups = "drop_last")%>% mutate(per =  100 *n_cells/sum(n_cells))
  
  integration_score = clusters_and_scores %>%  group_by(seurat_clusters) %>% filter(n_cells == max(n_cells)) %>% pull(per) %>% mean() %>% round(digits = 2)
  integration_score_sum_cells = clusters_and_scores %>%  group_by(seurat_clusters) %>% filter(n_cells == max(n_cells)) %>% pull(n_cells) %>% sum() %>%
    divide_by(sum(clusters_and_scores$n_cells)) %>% round(digits = 2)
  
  v_factor_levels <-c( "MesLike1", "MesLike2", "NPCLike1", "NPCLike2", "OPCLike","ACLike")
  colors = RColorBrewer::brewer.pal(6, "Paired"); colors[5] = "orange"
  p2 = ggplot(data=clusters_and_scores, aes(x=seurat_clusters, y=per, fill=factor(cancer_type, levels = v_factor_levels))) +
    geom_bar(stat="identity")+theme_minimal() + scale_fill_manual(values = colors,name  = "Cancer type")+ 
    labs(title = dataset@project.name,subtitle = "integration score=" %s+% integration_score %s+% "%" %s+% "\nintegration sums cells score="
         %s+% integration_score_sum_cells)+
    ylab("% from cluster")
  
  
  clusters_and_scores = FetchData(object = dataset,vars= c("orig.ident","seurat_clusters")) %>%  group_by(seurat_clusters,orig.ident) %>%  
    summarise(n_cells = n(), .groups = "drop_last")%>% mutate(per =  100 *n_cells/sum(n_cells))
  
  colors = RColorBrewer::brewer.pal(9, "Paired")
  p3 = ggplot(data=clusters_and_scores, aes(x=seurat_clusters, y=per, fill=factor(orig.ident))) +
    geom_bar(stat="identity")+theme_minimal() + scale_fill_manual(values = colors,name  = "Patient")+ 
    labs(title = dataset@project.name)+
    ylab("% from cluster")
  
  print_tab(p2+p3,title = dataset@project.name,subtitle_num = 3)
}

GBM_original

GBM_sipsic

GBM_sipsic_cp

GBM_seurat_integration

GBM_harmony

NA

4.3 Combine cancer subtypes

for (i in seq_along(all_datasets)) {
  all_datasets[[i]]$cancer_type_combined = all_datasets[[i]]$"cancer_type"  %>% gsub(pattern = "MesLike1|MesLike2",replacement = "MesLike")%>% gsub(pattern = "NPCLike1|NPCLike2",replacement = "NPCLike")
}

for (dataset in all_datasets) {
  clusters_and_scores = FetchData(object = dataset,vars= c("cancer_type_combined","seurat_clusters")) %>%  group_by(seurat_clusters,cancer_type_combined) %>%  summarise(n_cells = n(), .groups = "drop_last")%>% mutate(per =  100 *n_cells/sum(n_cells))
  
  integration_score = clusters_and_scores %>%  group_by(seurat_clusters) %>% filter(n_cells == max(n_cells)) %>% pull(per) %>% mean() %>% round(digits = 2)
  
  integration_score_sum_cells = clusters_and_scores %>%  group_by(seurat_clusters) %>% filter(n_cells == max(n_cells)) %>% pull(n_cells) %>% sum() %>% 
    divide_by(sum(clusters_and_scores$n_cells)) %>% round(digits = 2)
  
  v_factor_levels <-c( "MesLike", "NPCLike", "OPCLike","ACLike")
  colors = RColorBrewer::brewer.pal(6, "Paired")[c(2,4,5,6)]; colors[3] = "orange"
  p4 = ggplot(data=clusters_and_scores, aes(x=seurat_clusters, y=per, fill=factor(cancer_type_combined, levels = v_factor_levels))) +
    geom_bar(stat="identity")+theme_minimal() + scale_fill_manual(values = colors,name  = "Cancer type")+ 
    labs(title = dataset@project.name,subtitle = "integration score=" %s+% integration_score %s+% "%" %s+% "\nintegration sums cells score=" %s+% 
           integration_score_sum_cells)+ylab("% from cluster")
  print_tab(p4,title = dataset@project.name,subtitle_num = 3)
}

GBM_original

GBM_sipsic

GBM_sipsic_cp

GBM_seurat_integration

GBM_harmony

NA