First commit

.gitignore

+.Rproj.user
+.Rhistory
+.RData
+.Ruserdata

scRNAseq_day2.Rmd

+---
+title: "scRNAseq_day2"
+author: "Julia Angerer"
+date: "2024-06-03"
+output: html_document
+---
+
+# Annotate cell types
+
+## Step 1. Load library and your seurat object
+
+```{r}
+library(SingleR)
+library(Seurat)
+library(celldex)
+library(dplyr)
+library(pheatmap)
+
+sobj <- readRDS("C:/Users/julia/Documents/Master Biologie/Bioinformatics for Biologists/data/bc_tutorial.rds")
+sobj
+```
+
+## Step 2. Prepare inputs for running automatic cell type annotation
+
+We need to extract the normalization data using GetAssayData()
+
+```{r}
+sobj_count <- GetAssayData(sobj, layer = "data")
+str(sobj_count)
+```
+
+Get the Blueprint and Encode reference data from celldex package:
+
+```{r}
+ref <- celldex::BlueprintEncodeData()
+ref
+```
+
+## Step 3. Run SingleR to label cell type
+
+```{r}
+pred <- SingleR(test=sobj_count, ref=ref, labels=ref$label.main)
+pred %>% head()
+unique(pred$labels)
+```
+
+Next, we put the prediction results back to our seurat object for visualization
+
+```{r}
+sobj$labels <- pred[match(rownames(sobj@meta.data), rownames(pred)), 'labels']
+sobj@meta.data %>% head()
+```
+
+## Step 4. Visualize the data
+
+```{r}
+umap_pred <- DimPlot(
+  sobj, reduction = 'umap', group.by = 'labels', label = TRUE)
+
+umap_pred
+```
+
+Compare with your previous data
+
+```{r}
+umap_sobj <- DimPlot(
+  sobj, reduction = 'umap', group.by = 'seurat_clusters', label = TRUE)
+
+umap_sobj | umap_pred
+```
+
+Plot the SingleR assignment score accross all cell-label combination
+
+```{r}
+plotScoreHeatmap(pred)
+library(viridis)
+```
+
+## Step 5. Marker-based method: CellMarker v2
+
+Load the human markers from CellMarker 2.0, file includes cell markers of different cell types from different tissue in human
+
+```{r}
+human_marker <- read.csv("C:/Users/julia/Documents/Master Biologie/Bioinformatics for Biologists/data/CellMakerv2_Human.csv")
+str(human_marker)
+```
+
+Select marker for ovary from the tissue class column
+
+```{r}
+ovary_marker <- human_marker %>%
+  filter(species=="Human" & tissue_class=="Ovary") %>%
+  select(c(3,5,6,7,8,9,16))
+#select columns, also column name is possible
+```
+
+Get markers table from your seurat object
+
+```{r}
+sobj.markers <- FindAllMarkers(sobj, only.pos = TRUE)
+sobj.markers %>% head()
+```
+
+Combine CellMarker with ur markers sobj.markers
+
+```{r}
+merged_cellmarker <- sobj.markers %>%
+  filter(p_val_adj<0.05 & avg_log2FC>1 & pct.1 > 0.8 & pct.2 < 0.5) %>% 
+  left_join(ovary_marker[,c(3,4,6)], join_by(gene==marker), relationship = "many-to-many")
+            
+merged_cellmarker %>% head()
+```
+
+Let's try to see if there are fibroblasts and strocytes
+
+```{r}
+merged_cellmarker %>%
+  filter(cluster==2 & cell_type=="Cancer cell") %>%
+  select(c(7, 8, 9))
+```
+
+From the results, we found PDGFRA, CAV1, and ACTA2 poinitng to Fibroblasts in cancer cells
+
+Check again using FeaturePlot()
+
+```{r}
+genes <- c("PDGFRA", "CAV1", "ACTA2")
+FeaturePlot(sobj, features = genes)
+```
+
+## Step 6. Annotate the cells
+
+rename cluster 2 to Fibroblasts using RenameIdents()
+
+```{r}
+sobj_new <- RenameIdents(sobj, '2' = 'Fibroblasts')
+DimPlot(sobj_new, reduction = 'umap', label = TRUE)
+```
+
+## Step 7. Differentially expressed genes (Bonus)
+
+Can we distinguish cluster 7?
+
+```{r}
+merged_cellmarker %>%
+  filter(cluster==7 & cell_type=="Cancer cell") %>%
+  select(c(7,8,9)) %>%
+  str()
+```
+
+```{r}
+BiocManager::install("clusterProfiler")
+BiocManager::install("org.Hs.eg.db")
+BiocManager::install("DOSE")
+
+library(clusterProfiler)
+library(org.Hs.eg.db)
+library(DOSE)
+```
+
+```{r}
+marker_7_2 <- FindMarkers(sobj, ident.1 = 7, ident.2 = 2)
+marker_7_2_sig <- marker_7_2 %>%
+  filter(abs(avg_log2FC)>=1 & p_val_adj<0.05 & pct.2<0.4)
+str(marker_7_2_sig)
+```
+
+```{r}
+log2 <- marker_7_2_sig$avg_log2FC
+names(log2) <- rownames(marker_7_2_sig)
+
+log2_sort <- sort(log2, decreasing = TRUE)
+log2_sort[1:5]
+```
+
+Run gene set enrichment analysis:
+
+```{r}
+gse <- gseGO(geneList = log2_sort,
+             ont="ALL",
+             keyType="SYMBOL",
+             nPerm=10000,
+             OrgDb = "org.Hs.eg.db")
+```
+
+Visualize the results
+
+```{r}
+dotplot(gse)
+```
+
+Put the identity back
+
+```{r}
+sobj_new <- RenameIdents(sobj,
+                         '7' = 'Mesenchymal Cells',
+                         '2' = 'Fibroblasts')
+DimPlot(sobj_new, reduction = 'umap', label = TRUE)
+```

scrnaseq_tutorial.Rproj

+Version: 1.0
+
+RestoreWorkspace: Default
+SaveWorkspace: Default
+AlwaysSaveHistory: Default
+
+EnableCodeIndexing: Yes
+UseSpacesForTab: Yes
+NumSpacesForTab: 2
+Encoding: UTF-8
+
+RnwWeave: Sweave
+LaTeX: pdfLaTeX