Add Zhao data

code/prep-zhao.R

@@ -15,6 +15,23 @@ set.seed(07122021)
 
 zhao <- readRDS(args$raw_data)
 
-# Filter to only B cells
+# Remove cells without annotations.
+zhao <- zhao[, !is.na(zhao$fine)]
 
-# ...
+rownames(zhao) <- rowData(zhao)$Symbol
+
+clusters <- quickCluster(zhao)
+zhao <- computeSumFactors(zhao, clusters = clusters)
+zhao <- logNormCounts(zhao)
+
+dec_zhao <- modelGeneVarByPoisson(zhao)
+top_zhao <- getTopHVGs(dec_zhao, n = 2000)
+zhao <- zhao[top_zhao, ]
+
+zhao <- runPCA(zhao)
+zhao <- runTSNE(zhao, dimred = "PCA")
+zhao <- runUMAP(zhao, dimred = "PCA")
+
+colLabels(zhao) <- factor(zhao$fine)
+
+saveRDS(zhao, args$data)

config.yaml

 R: "/opt/R/4.1.0/bin/R"
 
 mg_type: ["umg", "sumg"]
-all_data_ids: ["zeisel", "pbmc3k", "lawlor", "endothelial", "astrocyte", "paul"]
+all_data_ids: ["zeisel", "pbmc3k", "lawlor", "endothelial", "astrocyte", "paul", "zhao"]
 general_sim_data_ids: ["zeisel", "pbmc3k", "lawlor", "paul"]
 time_sim_data_ids: ["pbmc3k"]