Update SMART-seq3 dataset preperation

code/prep-ss3_pbmc.R

@@ -31,7 +31,7 @@ ss3_pbmc <- computeSumFactors(ss3_pbmc, clusters = clusters)
 ss3_pbmc <- logNormCounts(ss3_pbmc)
 
 dec_ss3_pbmc <- modelGeneVarByPoisson(ss3_pbmc)
-top_ss3_pbmc <- getTopHVGs(dec_ss3_pbmc, n = 2000)
+top_ss3_pbmc <- getTopHVGs(dec_ss3_pbmc, n = 10000)
 ss3_pbmc <- ss3_pbmc[top_ss3_pbmc, ]
 
 ss3_pbmc <- runPCA(ss3_pbmc)
@@ -43,6 +43,9 @@ no_symbol_ind <- which(is.na(gene_name))
 gene_name[no_symbol_ind] <- rownames(ss3_pbmc)[no_symbol_ind]
 rownames(ss3_pbmc) <- gene_name
 
+dup_ind <- which(duplicated(rownames(ss3_pbmc)))
+ss3_pbmc <- ss3_pbmc[-dup_ind, ]
+
 colLabels(ss3_pbmc) <- factor(ss3_pbmc$clusterName)
 
 saveRDS(ss3_pbmc, args$data)