Ensure expert marker genes are preserved in the blood datasets

code/prep-dominguez_conde.R

@@ -19,9 +19,16 @@ clusters <- quickCluster(dominguez_conde)
 dominguez_conde <- computeSumFactors(dominguez_conde, clusters = clusters)
 dominguez_conde <- logNormCounts(dominguez_conde)
 
+blood_expert_mgs_info <- readRDS(
+  here::here("data", "expert_mgs", "blood_expert_mgs.rds")
+)
+dominguez_conde_ind <- blood_expert_mgs_info$cluster %in% dominguez_conde$Curated_annotation
+dominguez_conde_expert_mgs_info <- blood_expert_mgs_info[dominguez_conde_ind, ]
+dominguez_conde_expert_mgs <- unlist(dominguez_conde_expert_mgs_info$expert_mgs)
+
 dec_dominguez_conde <- modelGeneVarByPoisson(dominguez_conde)
 top_dominguez_conde <- getTopHVGs(dec_dominguez_conde, n = 2000)
-dominguez_conde <- dominguez_conde[top_dominguez_conde, ]
+dominguez_conde <- dominguez_conde[union(top_dominguez_conde, dominguez_conde_expert_mgs), ]
 
 dominguez_conde <- runPCA(dominguez_conde)
 dominguez_conde <- runTSNE(dominguez_conde, dimred = "PCA")

code/prep-ren.R

@@ -21,9 +21,16 @@ clusters <- quickCluster(ren)
 ren <- computeSumFactors(ren, clusters = clusters)
 ren <- logNormCounts(ren)
 
+blood_expert_mgs_info <- readRDS(
+  here::here("data", "expert_mgs", "blood_expert_mgs.rds")
+)
+ren_ind <- blood_expert_mgs_info$cluster %in% ren$Curated_annotation
+ren_expert_mgs_info <- blood_expert_mgs_info[ren_ind, ]
+ren_expert_mgs <- unlist(ren_expert_mgs_info$expert_mgs)
+
 dec_ren <- modelGeneVarByPoisson(ren)
 top_ren <- getTopHVGs(dec_ren, n = 2000)
-ren <- ren[top_ren, ]
+ren <- ren[union(top_ren, ren_expert_mgs), ]
 
 ren <- runPCA(ren)
 ren <- runTSNE(ren, dimred = "PCA")

code/prep-stephenson.R

@@ -21,9 +21,16 @@ clusters <- quickCluster(stephenson)
 stephenson <- computeSumFactors(stephenson, clusters = clusters)
 stephenson <- logNormCounts(stephenson)
 
+blood_expert_mgs_info <- readRDS(
+  here::here("data", "expert_mgs", "blood_expert_mgs.rds")
+)
+stephenson_ind <- blood_expert_mgs_info$cluster %in% stephenson$Curated_annotation
+stephenson_expert_mgs_info <- blood_expert_mgs_info[stephenson_ind, ]
+stephenson_expert_mgs <- unlist(stephenson_expert_mgs_info$expert_mgs)
+
 dec_stephenson <- modelGeneVarByPoisson(stephenson)
 top_stephenson <- getTopHVGs(dec_stephenson, n = 2000)
-stephenson <- stephenson[top_stephenson, ]
+stephenson <- stephenson[union(top_stephenson, stephenson_expert_mgs), ]
 
 stephenson <- runPCA(stephenson)
 stephenson <- runTSNE(stephenson, dimred = "PCA")

code/prep-yoshida.R

@@ -19,9 +19,16 @@ clusters <- quickCluster(yoshida)
 yoshida <- computeSumFactors(yoshida, clusters = clusters)
 yoshida <- logNormCounts(yoshida)
 
+blood_expert_mgs_info <- readRDS(
+  here::here("data", "expert_mgs", "blood_expert_mgs.rds")
+)
+yoshida_ind <- blood_expert_mgs_info$cluster %in% yoshida$Curated_annotation
+yoshida_expert_mgs_info <- blood_expert_mgs_info[yoshida_ind, ]
+yoshida_expert_mgs <- unlist(yoshida_expert_mgs_info$expert_mgs)
+
 dec_yoshida <- modelGeneVarByPoisson(yoshida)
 top_yoshida <- getTopHVGs(dec_yoshida, n = 2000)
-yoshida <- yoshida[top_yoshida, ]
+yoshida <- yoshida[union(top_yoshida, yoshida_expert_mgs), ]
 
 yoshida <- runPCA(yoshida)
 yoshida <- runTSNE(yoshida, dimred = "PCA")