gene_x 0 like s 489 view s
Tags: R, RNA-seq
单纯疱疹性脑炎是由单纯疱疹病毒（HSVs）引起的中枢神经系统的致命疾病。在使用抗病毒药物阿昔洛韦进行标准治疗后，大多数患者仍然出现各种神经后遗症。在这里，我们通过结合单细胞RNA测序、电生理和免疫染色来描述人脑器官样本中的HSV-1感染。我们观察到组织完整性、神经元功能和细胞转录组的强烈扰动。在阿昔洛韦治疗下，病毒复制被停止，但并未防止HSV-1引发的缺陷，如神经元过程和神经上皮的损伤。对感染后失调路径的无偏分析揭示了肿瘤坏死因子激活可能是一个潜在的致因因素。结合抗炎药物如necrostatin-1或bardoxolone methyl和抗病毒治疗可以防止感染引起的损害，这表明在急性感染中调整炎症反应可能会改进当前的治疗策略。
https://github.com/rajewsky-lab/HSV1_cerebral_organoids/blob/main/sc_and_bulkRNAseq_analysis.Rmd
---
title: "Code availability for the paper Modelling viral encephalitis caused by herpes simplex virus 1 infection in cerebral organoids"
output: html_notebook
---
```{r}
library(Seurat)
library(ggplot2)
library(pheatmap)
library(dplyr)
library(tidyr)
library(Matrix)
#library(Nebulosa)
library(edgeR)
library(clusterProfiler)
library(msigdbr)
library(AUCell)
library(DESeq2)
library(gprofiler2)
library(RColorBrewer)
options(bitmapType='cairo')
```
# Analysis of control (CTRL) organoids
Read digital gene expression matrices (generated with Spacemake) for CTRL samples and create Seurat objects for each sample keeping barcodes with at least 250 detected genes and genes detected at least 5 cells
```{r eval=FALSE, include=FALSE}
Xmoo1_CTRL_1 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_CTRL_1.txt.gz', sep = '', row.names = 1, header = T) %>%
CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_CTRL_1', assay = 'RNA')
Xmoo1_CTRL_1$line = 'iPSC line 1'
Xmoo1_CTRL_1$replicate = 'Rep. 1'
Xmoo1_CTRL_1$condition = 'CTRL'
Xmoo1_CTRL_2 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_CTRL_2.txt.gz', sep = '', row.names = 1, header = T) %>%
CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_CTRL_2', assay = 'RNA')
Xmoo1_CTRL_2$line = 'iPSC line 1'
Xmoo1_CTRL_2$replicate = 'Rep. 2'
Xmoo1_CTRL_2$condition = 'CTRL'
Gline_CTRL_1 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Gline_CTRL_1.txt.gz', sep = '', row.names = 1, header = T) %>%
CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Gline_CTRL_1', assay = 'RNA')
Gline_CTRL_1$line = 'iPSC line 2'
Gline_CTRL_1$replicate = 'Rep. 1'
Gline_CTRL_1$condition = 'CTRL'
Gline_CTRL_2 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Gline_CTRL_2.txt.gz', sep = '', row.names = 1, header = T) %>%
CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Gline_CTRL_2', assay = 'RNA')
Gline_CTRL_2$line = 'iPSC line 2'
Gline_CTRL_2$replicate = 'Rep. 2'
Gline_CTRL_2$condition = 'CTRL'
```
Create a single integrated Seurat object with all CTRL samples (2 replicates for 2 cell lines) after correcting for sample- and cell line-specific batch effects
```{r eval=FALSE, include=FALSE}
integration.list <- list(Xmoo1_CTRL_1, Xmoo1_CTRL_2, Gline_CTRL_1, Gline_CTRL_2)
for (i in 1:length(integration.list)) {
integration.list[[i]] <- SCTransform(integration.list[[i]], verbose = T,method= 'glmGamPoi', assay = 'RNA', new.assay.name = 'SCT', vst.flavor='v2')
}
integration.features <- SelectIntegrationFeatures(object.list = integration.list, nfeatures = 3000)
integration.list <- PrepSCTIntegration(object.list = integration.list, anchor.features = integration.features, verbose = T)
integration.anchors <- FindIntegrationAnchors(object.list = integration.list, normalization.method = "SCT", anchor.features = integration.features, verbose = T)
CTRL <- IntegrateData(anchorset = integration.anchors, normalization.method = "SCT", verbose = T)
DefaultAssay(CTRL) <- 'integrated'
remove(i, integration.list, integration.features, integration.anchors)
```
Perform dimensionality reduction and clustering with 30 principal components and default parameters
```{r eval=FALSE, include=FALSE}
CTRL <- CTRL %>% RunPCA( verbose = T) %>% RunUMAP( dims = 1:30, verbose = T) %>% FindNeighbors( dims = 1:30, verbose = T) %>% FindClusters( verbose = T)
```
Inspect transcript counts (nCount) and % of mitochondrial transcripts (percent_MT) to identify low quality clusters
```{r eval=FALSE, include=FALSE}
DefaultAssay(CTRL)= 'RNA'
CTRL= CTRL %>%PercentageFeatureSet( pattern = "^MT-", col.name = "percent_MT")
VlnPlot(CTRL, 'nCount_RNA', y.max = 2000, pt.size = 0)+ NoLegend()
VlnPlot(CTRL, 'percent_MT', pt.size = 0)+ NoLegend()
```
Remove low quality clusters 1 (low nCount) and 10 (low nCount, high mito) and select top 75% cells by nCount per cluster
```{r eval=FALSE, include=FALSE}
CTRL$barcode= colnames(CTRL)
top75= CTRL@meta.data[!CTRL$seurat_clusters %in% c(1, 10),] %>% group_by(seurat_clusters) %>% top_frac(n = 0.75, wt = nCount_RNA_exonic)
CTRL = subset(CTRL, cells=top75$barcode)
```
Re-run integration, dimensionality reduction and clustering
```{r eval=FALSE, include=FALSE}
Idents(CTRL) = 'sample'
integration.list <- SplitObject(CTRL)
for (i in 1:length(integration.list)) {
integration.list[[i]] <- SCTransform(integration.list[[i]],method= 'glmGamPoi', assay = 'RNA', new.assay.name = 'SCT', vst.flavor='v2')
}
integration.features <- SelectIntegrationFeatures(object.list = integration.list, nfeatures = 3000)
integration.list <- PrepSCTIntegration(object.list = integration.list, anchor.features = integration.features)
integration.anchors <- FindIntegrationAnchors(object.list = integration.list, normalization.method = "SCT", anchor.features = integration.features)
CTRL <- IntegrateData(anchorset = integration.anchors, normalization.method = "SCT")
DefaultAssay(CTRL) <- 'integrated'
CTRL <- CTRL %>% RunPCA() %>% RunUMAP( dims = 1:30) %>% FindNeighbors( dims = 1:30) %>% FindClusters(resolution = 0.4)
remove(i, integration.list, integration.features, integration.anchors)
```
Read in processed object
```{r}
CTRL= readRDS('seurat_objects/CTRL.rds')
```
## Extended figure 1.c
```{r}
DimPlot(CTRL, split.by = 'line', group.by = 'replicate', shuffle = T)
```
## Extended figure 1.d
```{r}
FeaturePlot(CTRL, features = c( 'VIM', 'SOX2', 'FOXG1','MKI67', 'DCX', 'EOMES', 'NEUROD2', 'NEUROD6', 'BCL11B', 'TCF7L2', 'GFAP', 'SPARCL1', 'RSPO2','RSPO3', 'TTR', 'DCT', 'SFRP2','COL1A2', 'RELN' ), slot = 'data', max.cutoff = 2, ncol = 5) * NoLegend() * NoAxes()
```
Important and prepare downloaded reference data (REF: Kanton et al, Nature 2019)
```{r eval=FALSE, include=FALSE}
# Read in cell-level metadata
kanton_meta = read.table('Kanton_et_al_2019/metadata_human_cells.tsv', sep = '\t', header = T)
rownames(kanton_meta)= paste0(kanton_meta$Sample, '_',kanton_meta$Barcode)
# Read in gene names
kanton_gene = read.table('Kanton_et_al_2019/genes_GRCh38.txt', sep = '\t', header = F)
# Read in gene expression matrix
kanton_mtx = readMM('Kanton_et_al_2019/human_cell_counts_GRCh38.mtx')
colnames(kanton_mtx) = rownames(kanton_meta)
rownames(kanton_mtx) = kanton_gene$V2
# Create Seurat object and perform SCT normalization
kanton = CreateSeuratObject(counts = kanton_mtx, meta.data = kanton_meta, min.cells = 50, min.features = 3) %>% SCTransform()
remove(kanton_meta, kanton_gene, kanton_mtx)
```
Label transfer from reference to CTRL organoids
```{r eval=FALSE, include=FALSE}
DefaultAssay(CTRL) <- 'integrated'
obj.list= list(kanton, CTRL)
features <- SelectIntegrationFeatures(obj.list)
anchors <- FindTransferAnchors(reference = kanton, query = CTRL, normalization.method = "SCT", reference.assay = 'SCT', query.assay = 'integrated',reduction = 'pcaproject', dims = 1:20, features = features, nn.method= 'rann', eps = 0.5, verbose = T)
CTRL[["kanton_PredCellType"]] <- TransferData(anchorset = anchors, refdata = kanton$PredCellType, prediction.assay = TRUE, weight.reduction = 'pcaproject', dims = 1:20, eps = 0.5 )
CTRL[["kanton_FullLineage"]] <- TransferData(anchorset = anchors, refdata = kanton$cl_FullLineage, prediction.assay = TRUE, weight.reduction = 'pcaproject', dims = 1:20, eps = 0.5 )
remove(kanton, obj.list, features, anchors)
```
## Supplementary figure 1.a
```{r}
FeaturePlot(CTRL, c('cycling dorsal progenitors', 'RGCs 3','IPs and early cortical neurons', 'cortical neurons 1', 'midbrain/hindbrain cells', 'Astrocyte', 'Mural', 'retina progenitors', 'Choroid'), order = T) * NoLegend() * NoAxes()
```
Annotate clusters
```{r eval=FALSE, include=FALSE}
Idents(CTRL) <- 'seurat_clusters'
CTRL <- RenameIdents(
object = CTRL,
'0'= 'Radial Glia G1/S Phase',
'1'= 'Immature Cortical Neurons 1',
'2'= 'Immature Cortical Neurons 2', #need better annotation
'3'= 'Thalamic Neurons',
'4'= 'Intermediate Progenitors',
'5'= 'Radial Glia G2M Phase',
'6'= 'Astroglia',
'7'= 'Mature Cortical Neurons',
'8'= 'Progenitors', #need better annotation
'9'= 'Cortical Hem/Choroid Plexus',
'10'= 'Hindbrain Neurons 1',
'11'= 'Retinal Progenitors',
'12'= 'Hindbrain Neurons 2', #need better annotation
'13'= 'Mural Cells',
'14'= 'Retinal Pigmented Cells',
'15'= 'Cajal Retzius Neurons')
CTRL$legend = Idents(CTRL)
```
## Figure 1.d
```{r}
color_palette=c('Immature Cortical Neurons 1'="#00133E", 'Immature Cortical Neurons 2'="#16365C", 'Mature Cortical Neurons'="#2D597A", 'Hindbrain Neurons 1'="#447C98", 'Thalamic Neurons'="#5A9FB6", 'Hindbrain Neurons 2'="#71C2D4", 'Cajal Retzius Neurons'="#87E5F2", 'Radial Glia G2M Phase'='#b7245c', 'Progenitors'='#8f0034', 'Radial Glia G1/S Phase'='#de598c', 'Intermediate Progenitors'='#7D5AA5', 'Retinal Progenitors'='#f7c548', 'Retinal Pigmented Cells'='#cf9d20', 'Mural Cells'='#060300', 'Cortical Hem/Choroid Plexus'='#FA8405', 'Astroglia'='#808080')
DimPlot(CTRL, cols = color_palette)
```
# Analysis of control (CTRL), HSV1 1 day post infection (1dpi), 3 days post infection (3dpi) and acyclovir (ACV) treated organoids
Read digital gene expression matrices (generated with Spacemake) for CTRL samples and create Seurat objects for each sample keeping barcodes with at least 250 detected genes and genes detected at least 5 cells
```{r eval=FALSE, include=FALSE}
Xmoo1_inf1dpi_1 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_inf1dpi_1.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_inf1dpi_1', assay = 'RNA')
Xmoo1_inf1dpi_1$line = 'iPSC line 1'
Xmoo1_inf1dpi_1$replicate = 'Rep. 1'
Xmoo1_inf1dpi_1$condition = 'HSV-1 (1dpi)'
Xmoo1_inf1dpi_2 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_inf1dpi_2.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_inf1dpi_2', assay = 'RNA')
Xmoo1_inf1dpi_2$line = 'iPSC line 1'
Xmoo1_inf1dpi_2$replicate = 'Rep. 2'
Xmoo1_inf1dpi_2$condition = 'HSV-1 (1dpi)'
Xmoo1_inf3dpi_1 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_inf3dpi_1.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_inf3dpi_1', assay = 'RNA')
Xmoo1_inf3dpi_1$line = 'iPSC line 1'
Xmoo1_inf3dpi_1$replicate = 'Rep. 1'
Xmoo1_inf3dpi_1$condition = 'HSV-1 (3dpi)'
Xmoo1_inf3dpi_2 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_inf3dpi_2.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_inf3dpi_2', assay = 'RNA')
Xmoo1_inf3dpi_2$line = 'iPSC line 1'
Xmoo1_inf3dpi_2$replicate = 'Rep. 2'
Xmoo1_inf3dpi_2$condition = 'HSV-1 (3dpi)'
Xmoo1_inf3dpiACV_1 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_inf3dpiACV_1.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_inf3dpiACV_1', assay = 'RNA')
Xmoo1_inf3dpiACV_1$line = 'iPSC line 1'
Xmoo1_inf3dpiACV_1$replicate = 'Rep. 1'
Xmoo1_inf3dpiACV_1$condition = 'HSV-1 (3dpi) +ACV'
Xmoo1_inf3dpiACV_2 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_inf3dpiACV_2.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_inf3dpiACV_2', assay = 'RNA')
Xmoo1_inf3dpiACV_2$line = 'iPSC line 1'
Xmoo1_inf3dpiACV_2$replicate = 'Rep. 2'
Xmoo1_inf3dpiACV_2$condition = 'HSV-1 (3dpi)'
Gline_inf1dpi_1 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Gline_inf1dpi_1.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Gline_inf1dpi_1', assay = 'RNA')
Gline_inf1dpi_1$line = 'iPSC line 2'
Gline_inf1dpi_1$replicate = 'Rep. 1'
Gline_inf1dpi_1$condition = 'HSV-1 (1dpi)'
Gline_inf1dpi_2 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Gline_inf1dpi_2.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Gline_inf1dpi_2', assay = 'RNA')
Gline_inf1dpi_2$line = 'iPSC line 1'
Gline_inf1dpi_2$replicate = 'Rep. 2'
Gline_inf1dpi_2$condition = 'HSV-1 (1dpi)'
Gline_inf3dpi_1 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Gline_inf3dpi_1.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Gline_inf3dpi_1', assay = 'RNA')
Gline_inf3dpi_1$line = 'iPSC line 2'
Gline_inf3dpi_1$replicate = 'Rep. 1'
Gline_inf3dpi_1$condition = 'HSV-1 (3dpi)'
Gline_inf3dpi_2 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Gline_inf3dpi_2.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Gline_inf3dpi_2', assay = 'RNA')
Gline_inf3dpi_2$line = 'iPSC line 1'
Gline_inf3dpi_2$replicate = 'Rep. 2'
Gline_inf3dpi_2$condition = 'HSV-1 (3dpi)'
Gline_inf3dpiACV_1 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Gline_inf3dpiACV_1.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Gline_inf3dpiACV_1', assay = 'RNA')
Gline_inf3dpiACV_1$line = 'iPSC line 2'
Gline_inf3dpiACV_1$replicate = 'Rep. 1'
Gline_inf3dpiACV_1$condition = 'HSV-1 (3dpi) +ACV'
Gline_inf3dpiACV_2 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Gline_inf3dpiACV_2.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Gline_inf3dpiACV_2', assay = 'RNA')
Gline_inf3dpiACV_2$line = 'iPSC line 1'
Gline_inf3dpiACV_2$replicate = 'Rep. 2'
Gline_inf3dpiACV_2$condition = 'HSV-1 (3dpi) +ACV'
```
Create a single integrated Seurat object with all 16 samples (2 replicates for 2 cell lines per 4 conditions) after correcting for condition- and cell line-specific effects
```{r eval=FALSE, include=FALSE}
all_timepoints= merge(Xmoo1_CTRL_1_exonic, list(Xmoo1_CTRL_2_exonic, Xmoo1_inf1dpi_1, Xmoo1_inf1dpi_2, Xmoo1_inf3dpi_1, Xmoo1_inf3dpi_2, Xmoo1_inf3dpiACV_1, Xmoo1_inf3dpiACV_2,
Gline_CTRL_1_exonic, Gline_CTRL_2_exonic, Gline_inf1dpi_1, Gline_inf1dpi_2, Gline_inf3dpi_1, Gline_inf3dpi_2, Gline_inf3dpiACV_1, Gline_inf3dpiACV_2))
all_timepoints$line_condition= paste0(all_timepoints$line, '_', all_timepoints$condition)
Idents(all_timepoints) = 'line_condition'
integration.list <- SplitObject(all_timepoints)
for (i in 1:length(integration.list)) {
integration.list[[i]] <- SCTransform(integration.list[[i]], verbose = T,method= 'glmGamPoi', assay = 'RNA', new.assay.name = 'SCT', vst.flavor='v2')
}
integration.features <- SelectIntegrationFeatures(object.list = integration.list, nfeatures = 3000)
integration.list <- PrepSCTIntegration(object.list = integration.list, anchor.features = integration.features, verbose = T)
integration.anchors <- FindIntegrationAnchors(object.list = integration.list, normalization.method = "SCT", anchor.features = integration.features, verbose = T)
all_timepoints <- IntegrateData(anchorset = integration.anchors, normalization.method = "SCT", verbose = T)
DefaultAssay(all_timepoints) <- 'integrated'
remove(i, integration.list, integration.features, integration.anchors)
```
Perform dimensionality reduction and clustering with 30 principal components and default parameters
```{r eval=FALSE, include=FALSE}
all_timepoints <- all_timepoints %>% RunPCA( verbose = T) %>% RunUMAP( dims = 1:30, verbose = T) %>% FindNeighbors( dims = 1:30, verbose = T) %>% FindClusters( verbose = T)
```
Inspect transcript counts (nCount) and % of mitochondrial transcripts (percent_MT) to identify low quality clusters
```{r eval=FALSE, include=FALSE}
VlnPlot(all_timepoints, 'nCount_RNA_exonic', y.max = 2000, pt.size = 0)+ NoLegend()
VlnPlot(all_timepoints, 'percent_MT', y.max = 2000, pt.size = 0)+ NoLegend()
```
Remove low quality clusters 6 (low nCount) and 10 (high mito) and select top 75% cells by nCount per cluster
```{r eval=FALSE, include=FALSE}
all_timepoints$barcode= colnames(all_timepoints)
top75= all_timepoints@meta.data[!all_timepoints$seurat_clusters %in% c(6, 10),] %>% group_by(seurat_clusters) %>% top_frac(n = 0.75, wt = nCount_RNA_exonic)
all_timepoints = subset(all_timepoints, cells=top75$barcode)
```
Re-run integration, dimensionality reduction and clustering
```{r eval=FALSE, include=FALSE}
Idents(all_timepoints) = 'line_condition'
integration.list <- SplitObject(all_timepoints)
for (i in 1:length(integration.list)) {
integration.list[[i]] <- SCTransform(integration.list[[i]],method= 'glmGamPoi', assay = 'RNA', new.assay.name = 'SCT', vst.flavor='v2')
}
integration.features <- SelectIntegrationFeatures(object.list = integration.list, nfeatures = 3000)
integration.list <- PrepSCTIntegration(object.list = integration.list, anchor.features = integration.features)
integration.anchors <- FindIntegrationAnchors(object.list = integration.list, normalization.method = "SCT", anchor.features = integration.features)
all_timepoints <- IntegrateData(anchorset = integration.anchors, normalization.method = "SCT")
DefaultAssay(all_timepoints) <- 'integrated'
all_timepoints <- all_timepoints %>% RunPCA() %>% RunUMAP( dims = 1:30) %>% FindNeighbors( dims = 1:30) %>% FindClusters(resolution = 0.8)
remove(i, integration.list, integration.features, integration.anchors)
```
Quantify the % of CTRL cells (condition) in each cluster
```{r eval=FALSE, include=FALSE}
round(prop.table(table(all_timepoints$seurat_clusters, all_timepoints$condition=='CTRL'), margin=1)*100, digits=1)
```
Add annotations of CTRL cells
```{r eval=FALSE, include=FALSE}
all_timepoints$legend_CTRL = CTRL$legend
```
Quantify the % of CTRL cell type (legend) in each cluster
```{r eval=FALSE, include=FALSE}
round(prop.table(table(all_timepoints$legend_CTRL, all_timepoints$seurat_clusters), margin=1)*100, digits=1)
```
Identify marker genes for each cluster
```{r eval=FALSE, include=FALSE}
all_timepoints_markers= FindAllMarkers(all_timepoints, only.pos = T)
```
Cluster annotation strategy:
1. Clusters composed by at least 33% of CTRL cells (as CTRL cell represent 33% of the dataset) take the annotation of the majority of the CTRL cells
0: Mixed Neurons (35+31%)
3: Cortical Neurons (56 Immature, 34 Mature %)
4: RG (89%)
5: Astroglia (68.5%)
8: RG (76.5%)
12: IPC (86%)
13: RG proliferating (97%)
15: Thalamic (54% Thalamic)
18: Mural (95%)
2. Clusters that contain more than than 50% of the CTRL cell types from a specific cluster and at least 10% of CTRL cells
7: Retinal Progenitors (90%) [10% CTRL]
11: Thalamic Neurons (54%) [26% CTRL]
14: Choroid (94%) [28% CTRL]
16: Hindbrain (65%) [22% CTRL]
17: Retinal Pigmented (81%) [22% CTRL]
3. Missing clusters annotated on markers
1: Infected
2: Infected
6: Infected
9: Infected
10: Hindbrain (HOX genes: HOXA5, B3-5-6)
```{r include=FALSE}
Idents(all_timepoints) <- 'seurat_clusters'
all_timepoints <- RenameIdents(
object = all_timepoints,
'0'= 'Mixed Neurons',
'1'= 'Highly infected 1',
'2'= 'Highly infected 2',
'3'= 'Cortical Neurons',
'4'= 'Radial Glia G1 Phase',
'5'= 'Astroglia',
'6'= 'Highly infected 3',
'7'= 'Retinal Progenitors',
'8'= 'Radial Glia S Phase',
'9'= 'Highly infected 4',
'10'= 'Hindbrain Neurons 1',
'11'= 'Thalamic Neurons 1',
'12'= 'Intermediate Progenitors',
'13'= 'Radial Glia G2M Phase',
'14'= 'Cortical Hem/Choroid Plexus',
'15'= 'Thalamic Neurons 2',
'16'= 'Hindbrain Neurons 2',
'17'= 'Retinal Pigmented Cells',
'18'= 'Mural Cells')
all_timepoints$legend = Idents(all_timepoints)
```
CTRL cells in infected clusters have lower UMIs than CTRL cells in other clusters
```{r eval=FALSE, include=FALSE}
ggplot(all_timepoints@meta.data[all_timepoints$condition=='CTRL',], aes(x= legend, y= nCount_RNA)) + geom_boxplot() + ylim(c(0, 3000)) + RotatedAxis()
```
CTRL cells in infected clusters have lower UMIs than infected cells in Highly infected clusters
```{r eval=FALSE, include=FALSE}
ggplot(all_timepoints@meta.data, aes(x= seurat_clusters, y= nCount_RNA, fill=condition)) + geom_boxplot() + ylim(c(0, 3000))
```
Removing CTRL cells from 'Highly infected' clusters
```{r eval=FALSE, include=FALSE}
all_timepoints =subset(all_timepoints, cells = colnames(all_timepoints)[!(all_timepoints$condition== 'CTRL' & all_timepoints$seurat_clusters %in% c(1, 2, 6, 9) )])
```
Import processed object
```{r}
saveRDS(all_timepoints, 'seurat_objects/all_timepoints.rds')
```
## Figure 2.a
```{r}
color_palette=c( 'Mixed Neurons'="#00133E", 'Thalamic Neurons 2'="#16365C",'Cortical Neurons'="#2D597A",'Hindbrain Neurons 1'="#447C98",'Thalamic Neurons 1'="#5A9FB6",'Hindbrain Neurons 2'="#71C2D4",'Radial Glia G2M Phase'='#b7245c','Radial Glia S Phase'='#8f0034', 'Radial Glia G1 Phase'='#de598c','Intermediate Progenitors'='#7D5AA5','Retinal Progenitors'='#f7c548','Retinal Pigmented Cells'='#cf9d20','Mural Cells'='#060300','Cortical Hem/Choroid Plexus'='#FA8405','Astroglia'='808080','Highly infected 1'='#23A17B','Highly infected 2'='#17B67E','Highly infected 3'='#0CCA80','Highly infected 4'='#005225')
DimPlot(all_timepoints, cols = color_palette)
```
## Extended figure 3.c
```{r}
DimPlot(all_timepoints, split.by = 'line_condition', group.by = 'replicate', shuffle = T, ncol = 4)
```
## Figure 3.b
```{r}
all_timepoints$legend= factor(all_timepoints$legend, levels = names(color_palette))
ggplot(all_timepoints@meta.data, aes(x=condition, fill=legend)) +
geom_bar(position='fill') +
scale_fill_manual(values=color_palette) +
theme_bw() +
xlab('') + ylab('Cluster %') +
scale_y_continuous(labels = scales::percent_format())
```
```{r}
extended_3b= all_timepoints@meta.data %>% group_by(condition, legend) %>% summarise(count= length(condition)) %>% pivot_wider(names_from = legend, values_from = count, values_fill = 0) %>% as.data.frame()
rownames(extended_3b)= extended_3b$condition
extended_3b$condition= NULL
write.csv(round(extended_3b/rowSums(extended_3b)*100, digits = 1), '../gene_lists/source_extended_Figure_3b.csv')
```
## Extended figure 4.a
```{r}
all_timepoints$sample= paste0(all_timepoints$condition, '_', all_timepoints$replicate)
ggplot(all_timepoints@meta.data, aes(x=sample, fill=legend)) +
geom_bar(position='fill') + scale_fill_manual(values=color_palette) +
theme_bw() +
xlab('') + ylab('Cluster %') +
scale_y_continuous(labels = scales::percent_format()) + facet_wrap('line')
```
Compute viral load (% of viral transcripts) per cell
```{r eval=FALSE, include=FALSE}
# Identify viral genes
viral_genes= grep("^US[0-9]{1,2}A{0,1}[0-9]{0,1}|^UL[0-9]{1,2}[A]{0,1}[0-9]{0,1}|^R[LS][12]|^LAT|^GFP$",
rownames(all_timepoints), value="TRUE", perl=TRUE)
# Extract viral gene expression
viral_expression= all_timepoints@assays$RNA[viral_genes, ]
# Count viral transcripts per cell
all_timepoints$nCount_viral= colSums(viral_expression)
# Compute viral load
all_timepoints$viral_load = all_timepoints$nCount_viral/all_timepoints$nCount_RNA*100
```
## Figure 3.c
```{r}
viral_load_matrix= all_timepoints@meta.data %>% group_by(legend, condition) %>% summarise(viral_load= round(median(viral_load), digits = 1) ) %>% tidyr::pivot_wider(names_from = condition, values_from = viral_load) %>% as.data.frame()
rownames(viral_load_matrix) = as.character(viral_load_matrix$legend)
viral_load_matrix$legend= NULL
green_palette= c(colorRampPalette(c("white", "#23A17B"))(10), rep("#23A17B", 60))
pheatmap(viral_load_matrix, cluster_rows = F, cluster_cols = F, display_numbers = T, fontsize_number = 4, number_format = "%.1f",color = green_palette, angle_col = 90, labels_col = c('CTRL', '+HSV-1 1dpi', '+HSV-1 3dpi', '+HSV-1 3dpi +ACV'), na_col = 'white', border_color = 'grey', cellwidth = 12, cellheight = 12, show_rownames = F, main = 'Median viral \n transcript %', fontsize = 10, fontsize_col = 10)
```
## Extended figure 4.c
```{r}
viral_metadata= all_timepoints@meta.data[, c('viral_load', 'condition','legend')]
viral_metadata= viral_metadata[order(viral_metadata$condition, viral_metadata$legend ),]
viral_expression= log10(viral_expression[order(rowSums(viral_expression), decreasing = T) , rownames(viral_metadata)]+1)
condition_cols= c('CTRL'='grey', 'HSV-1 (1dpi)'='#F1B3F4', 'HSV-1 (3dpi)'='#A13FA6', 'HSV-1 (3dpi) +ACV'='#BE64C2')
pheatmap(viral_expression, cluster_rows = F, cluster_cols = F, display_numbers = F, color = colorRampPalette(c("white", "#23A17B"))(100), angle_col = 90, show_rownames = T, main = 'Viral gene expression (log10)', fontsize = 10, fontsize_col = 10, annotation_col = viral_metadata, annotation_colors = list(legend= color_palette, condition= condition_cols ), show_colnames = F)
```
Create a Seurat object only with 'Highly infected' cells
```{r}
infected_clusters= subset(all_timepoints, cells= colnames(all_timepoints)[grep('Highly infected' ,all_timepoints$legend)])
DefaultAssay(infected_clusters) = 'SCT'
```
## Extended figure 4.d
```{r}
ggplot(infected_clusters@meta.data, aes(x=legend, fill= condition))+geom_bar(position = 'fill') +
scale_fill_manual(values = c( '#F1B3F4', '#A13FA6', '#BE64C2')) +
xlab('') + ylab('Condition %') +
scale_y_continuous(labels = scales::percent_format())
```
## Extended figure 4.e
```{r}
DotPlot(infected_clusters , features = c( 'VIM', 'SOX2', 'FOXG1','MKI67', 'DCX', 'EOMES', 'NEUROD2', 'NEUROD6', 'BCL11B', 'TCF7L2' ,'GFAP', 'SPARCL1', 'RSPO2', 'RSPO3', 'TTR', 'DCT', 'SFRP2','COL1A2','RELN'), group.by = 'legend', scale=T )
```
## Extended figure 4.f
```{r}
VlnPlot(infected_clusters, features = c('viral_load'), group.by = 'legend', cols =color_palette) + NoLegend() + ggtitle('') + ylab('Viral load (% UMI)') + xlab('')
remove(infected_clusters)
```
Store condition as a 1-hot encoded assay
```{r}
sample_assay = pivot_wider(data.frame(condition=all_timepoints$condition, values= 1, cells= colnames(all_timepoints)), names_from = 'condition', values_from = values, values_fill = 0 )
sample_assay = data.frame(sample_assay[, 2:5], row.names = sample_assay$cells)
sample_assay= sample_assay/colSums(sample_assay)
all_timepoints[["condition_assay"]] = CreateAssayObject(counts = t(sample_assay) )
```
## Figure 3.d
```{r}
DefaultAssay(all_timepoints) = 'condition_assay'
plot_density(object = all_timepoints, features = rownames(all_timepoints) )
all_timepoints@assays$condition_assay=NULL
```
## Supplementary figure 2.a
```{r}
VlnPlot(all_timepoints, features ='STMN2', group.by = 'condition', cols = condition_cols, assay = 'RNA', log=T, slot = 'data') + NoLegend() +ggtitle('') + ylab('Log normalized counts')
```
Identify differentially expressed genes in each cluster for selected clusters
```{r}
# Prepare dataframe to store results
DE_genes = data.frame(logFC= NA, logCPM= NA,F= NA, PValue = NA, cluster=NA, gene =NA, contrast= NA, adjPvalue= NA )
# Set sample name order
all_timepoints$line_condition_replicate = paste0(all_timepoints$line, '_', all_timepoints$condition, '_', all_timepoints$replicate)
all_timepoints$line_condition_replicate = factor(all_timepoints$line_condition_replicate, levels= c('iPSC line 2_CTRL_Rep. 1', 'iPSC line 2_CTRL_Rep. 2','iPSC line 1_CTRL_Rep. 1', 'iPSC line 1_CTRL_Rep. 2',
'iPSC line 2_HSV-1 (1dpi)_Rep. 1', 'iPSC line 2_HSV-1 (1dpi)_Rep. 2', 'iPSC line 1_HSV-1 (1dpi)_Rep. 1', 'iPSC line 1_HSV-1 (1dpi)_Rep. 2',
'iPSC line 2_HSV-1 (3dpi)_Rep. 1', 'iPSC line 2_HSV-1 (3dpi)_Rep. 2', 'iPSC line 1_HSV-1 (3dpi)_Rep. 1', 'iPSC line 1_HSV-1 (3dpi)_Rep. 2',
'iPSC line 2_HSV-1 (3dpi) +ACV_Rep. 1', 'iPSC line 2_HSV-1 (3dpi) +ACV_Rep. 2', 'iPSC line 1_HSV-1 (3dpi) +ACV_Rep. 1', 'iPSC line 1_HSV-1 (3dpi) +ACV_Rep. 2'))
all_timepoints$barcode= colnames(all_timepoints)
for (i in levels(all_timepoints$legend)[c(1,3, 5, 9)] ){
# Subsample to 100 cells per organoid per cluster to correct for different abundances across conditions
random_100_subset = all_timepoints@meta.data[all_timepoints$legend == i, ] %>% group_by(line_condition_replicate) %>% sample_n(100, replace= T)
cells= random_100_subset$barcode
cluster_group= random_100_subset$line_condition_replicate
# Sum gene expression across cells in each organoid to generate a pseudobulk gene expression matrix
pseudobulk= t(rowsum(x = t(as.matrix(all_timepoints@assays$RNA@counts[, cells])), group = cluster_group))
# Define sample metadata
condition = factor(c('CTRL','CTRL','CTRL','CTRL', 'inf1dpi','inf1dpi','inf1dpi','inf1dpi', 'inf3dpi', 'inf3dpi','inf3dpi', 'inf3dpi', 'inf3dpi_ACV','inf3dpi_ACV', 'inf3dpi_ACV','inf3dpi_ACV'),
levels= c('CTRL','inf1dpi', 'inf3dpi', 'inf3dpi_ACV'))
line= factor(c('Gline','Gline','Xmoo1','Xmoo1', 'Gline','Gline','Xmoo1','Xmoo1', 'Gline', 'Gline','Xmoo1', 'Xmoo1', 'Gline','Gline', 'Xmoo1','Xmoo1'),
levels= c('Gline','Xmoo1'))
# Define experimental design and remove cell-line specific effects
design = model.matrix(~line + condition)
# Define comparisons to assess
my.contrasts <- makeContrasts(
inf1dpivsCTRL = conditioninf1dpi,
inf3dpivsCTRL = conditioninf3dpi,
inf3dpi_ACVvsCTRL = conditioninf3dpi_ACV,
inf3dpivsinf1dpi= conditioninf3dpi - conditioninf1dpi,
inf3dpi_ACVvsinf3dpi= conditioninf3dpi_ACV - conditioninf3dpi,
inf3dpi_ACVvsinf1dpi= conditioninf3dpi_ACV - conditioninf1dpi,
levels = colnames(design))
# Create edgeR object
DE_object = DGEList(counts=pseudobulk,group=condition )
# Filter lowly detected, potentially noisy genes
DE_object = DE_object[filterByExpr(DE_object, min.count= 10, min.total.count= 50), , keep.lib.sizes=FALSE] # remove genes with >100 tpm in at least 2 samples
# Adjust by library size
DE_object$samples$lib.size <- colSums(DE_object$counts)
DE_object <- calcNormFactors(DE_object)
# Fit statistical model
DE_object = estimateDisp(DE_object, design)
DE_object = glmQLFit(DE_object, design)
# Prepare cluster-level dataframe to store results
cluster_DE_genes = data.frame(logFC= NA, logCPM= NA, F =NA, PValue = NA, cluster=NA, gene =NA, contrast= NA, adjPvalue= NA)
for (contrast in colnames(my.contrasts)){
# Compute differentially expressed genes
cluster_comparison= glmQLFTest(DE_object, contrast=my.contrasts[,contrast])$table
# Store metadata
cluster_comparison$cluster = i
cluster_comparison$gene = rownames(cluster_comparison)
cluster_comparison$contrast =contrast
# Compute bonferroni-adjusted p value
cluster_comparison$adjPvalue = p.adjust(cluster_comparison$PValue)
# Add to result data frame
cluster_DE_genes = rbind.data.frame(cluster_DE_genes, cluster_comparison)
cluster_DE_genes= cluster_DE_genes[!is.na(cluster_DE_genes$gene), ]
}
# Add to global result dataframe
DE_genes = rbind.data.frame(DE_genes, cluster_DE_genes)
}
# Exclude viral genes
DE_genes$viral = DE_genes$gene %in% viral_genes
DE_genes= DE_genes[!is.na(DE_genes$gene), ]
# Save results to file
write.csv(DE_genes, 'gene_lists/differentially_expressed_genes.csv')
```
Define function to run gene set enrichment analysis on differential expressed genes in each comparison and for each cluster ad save the results as separate RDS files
```{r}
run_GSEA_edgeR = function(DE_genes_csv, species = 'Homo sapiens', category, subcategory = NULL, cluster, contrasts, cluster_name,scoreType = "std") {
# Check if output directory exist and create it if it doesn't
if (file.exists(paste0('GSEA/', cluster_name))== F){dir.create(paste0('GSEA/', cluster_name))}
# Read in differential expressed genes
differential_genes = read.csv(DE_genes_csv)
# Define gene set database
msigdbr_t2g = msigdbr(species = species, category = category, subcategory = subcategory) %>% dplyr::distinct(gs_name, gene_symbol) %>% as.data.frame()
# Prepare list for storing results of each comparison
contrast_gsea = list()
for (contrast in contrasts) {
# Define differential genes and order them according to decreasing fold change
genes = differential_genes[differential_genes$contrast == contrast & differential_genes$cluster == cluster,]
genes = genes[order(genes$logFC, decreasing = T),]
genes_for_GSEA = genes$logFC
names(genes_for_GSEA) = genes$gene
# Compute gene set enrichments
if(any(genes$gene %in% msigdbr_t2g$gene_symbol)){contrast_gsea[as.character(contrast)] = GSEA(geneList = genes_for_GSEA, TERM2GENE = msigdbr_t2g, scoreType= scoreType)}
}
# Save file to disk
saveRDS(contrast_gsea, paste0('GSEA/', cluster_name,'/', category, '_', subcategory, '.rds'))
}
```
Run GSEA for the HALLMARK database
```{r}
dir.create('GSEA')
for (cluster in levels(all_timepoints$legend)[c(1,3, 5, 9)]){
run_GSEA_edgeR(DE_genes_csv ='gene_lists/differentially_expressed_genes.csv', category = 'H',
contrasts= c('inf1dpivsCTRL', '+HSV-1 3dpi vs CTRL', 'inf3dpi_ACVvsCTRL'),
cluster = cluster, cluster_name = cluster)}
```
Read in GSEA results
```{r}
GSEA_results = data.frame(ID= NA, NES= NA, p.adjust =NA, contrast = NA, cluster=NA)
for (cluster in levels(all_timepoints$legend)[c(1,3, 5, 9)]){
cluster_rds = readRDS(paste0('GSEA/', cluster,'/H_.rds'))
cluster_enrichments= data.frame(ID= NA, NES= NA, p.adjust =NA, contrast = NA)
if(!is.null(cluster_rds$inf1dpivsCTRL)){ if(nrow(cluster_rds$inf1dpivsCTRL@result)>0){cluster_enrichments = rbind.data.frame(cluster_enrichments, data.frame(cluster_rds$inf1dpivsCTR@result[, c('ID', 'NES', 'p.adjust')], contrast= 'inf1dpivsCTRL'))}}
if(!is.null(cluster_rds$inf3dpivsCTRL)){if(nrow(cluster_rds$inf3dpivsCTRL@result)>0){cluster_enrichments = rbind.data.frame(cluster_enrichments, data.frame(cluster_rds$inf3dpivsCTR@result[, c('ID', 'NES', 'p.adjust')], contrast= 'inf3dpivsCTRL' ))}}
if(!is.null(cluster_rds$inf3dpi_ACVvsCTRL)){if(nrow(cluster_rds$inf3dpi_ACVvsCTRL@result)>0){cluster_enrichments = rbind.data.frame(cluster_enrichments, data.frame(cluster_rds$inf3dpi_ACVvsCTRL@result[, c('ID', 'NES', 'p.adjust')], contrast= 'inf3dpi_ACVvsCTRL' ))}}
cluster_enrichments$cluster= cluster
GSEA_results = rbind.data.frame(GSEA_results, cluster_enrichments)
}
# Clean and polish labels
GSEA_results= GSEA_results[!is.na(GSEA_results$ID), ]
GSEA_results$ID= gsub('HALLMARK_', '', GSEA_results$ID)
GSEA_results$ID= gsub('_', ' ', GSEA_results$ID)
GSEA_results$ID= gsub('_', ' ', GSEA_results$ID)
GSEA_results$ID= stringr::str_to_sentence(GSEA_results$ID)
GSEA_results$ID= gsub('nfkb', 'NFkB', GSEA_results$ID)
GSEA_results$ID= gsub('Tnfa', 'TNFa', GSEA_results$ID)
GSEA_results$ID= gsub('Uv', 'UV', GSEA_results$ID)
GSEA_results$ID= gsub('Mtorc', 'mTORC', GSEA_results$ID)
GSEA_results$ID= gsub('Dna', 'DNA', GSEA_results$ID)
```
## Figure 4.a
```{r}
# Define gene set order based on the most frequently up or down regulated clusters
gene_set_stats=GSEA_results %>% group_by(ID) %>% summarise(positive = sum(NES >0), negative= sum(NES<0)) %>% as.data.frame()
gene_set_stats= gene_set_stats[order( gene_set_stats$positive, -gene_set_stats$negative, decreasing = T ),]
GSEA_results$ID= factor(GSEA_results$ID, levels = rev(gene_set_stats$ID))
# Define cluster, gene set and comparison order
GSEA_results$contrast = factor(GSEA_results$contrast, levels= c('inf1dpivsCTRL', 'inf3dpivsCTRL', 'inf3dpi_ACVvsCTRL'))
GSEA_results$cluster= factor(GSEA_results$cluster, levels = c('Radial Glia G1 Phase', 'Cortical Neurons', 'Mixed Neurons', 'Thalamic Neurons 1'))
# Compute -log10 p values
GSEA_results$`-log10 adj. p-value`= -log10(GSEA_results$p.adjust)
ggplot(GSEA_results,
aes(x = contrast, y = ID, col = NES, size = `-log10 adj. p-value`)) +
geom_point() + scale_color_continuous(low = "#00BFC4", high = "#F8766D")+
scale_color_gradientn(colors = c("dodgerblue", "firebrick"))+
facet_wrap('cluster', ncol = 4) +
theme_bw() + RotatedAxis()+ xlab('') + ylab('')
```
## Extended figure 4.a
```{r}
# Retrieve all genes included in the 'TNFA SIGNALLING VIA NFKB' gene set
msigdbr_t2g = msigdbr(species = 'Homo sapiens', category = 'H', subcategory = NA) %>% dplyr::distinct(gs_name, gene_symbol) %>% as.data.frame()
tnfa_genes= msigdbr_t2g[msigdbr_t2g$gs_name == 'HALLMARK_TNFA_SIGNALING_VIA_NFKB',2]
# Keep only genes detected in the scRNAseq data
DefaultAssay(all_timepoints)= 'SCT'
tnfa_genes= tnfa_genes[tnfa_genes %in% rownames(all_timepoints)]
# Retrieve their average gene expression
all_timepoints$legend_condition= paste0(all_timepoints$legend,' ' ,all_timepoints$condition)
gene_expression= AverageExpression(subset(all_timepoints, cells = all_timepoints$barcode[all_timepoints$legend %in% levels(all_timepoints$legend)[c(1,3, 5, 9, 16:20)]]),
assays = 'SCT', slot = 'counts', features = tnfa_genes, group.by = 'legend_condition')
gene_expression= as.data.frame(gene_expression$SCT)
gene_expression$gene= rownames(gene_expression)
# Compute the percentage of expressing cells
pct_expression_data= data.frame(fake_cluster= 1:length(tnfa_genes), row.names = tnfa_genes )
for (cluster in colnames(gene_expression) ){
cells= all_timepoints$barcode[all_timepoints$legend_condition==cluster]
cluster_expression= as.data.frame(all_timepoints@assays$RNA@counts[tnfa_genes, cells]>0)
pct_expression= data.frame(apply(cluster_expression, MARGIN = 1, FUN = function(x){ round(sum(x)/length(x)*100, digits = 0) } ))
colnames(pct_expression)= cluster
pct_expression_data= cbind(pct_expression_data, pct_expression)
}
remove(cluster, pct_expression)
pct_expression_data=pct_expression_data[, 2:29]
pct_expression_data$gene= rownames(pct_expression_data)
pct_expression_data = tidyr::pivot_longer(pct_expression_data, cols = 1:28, values_to = 'pct_expression', names_to = 'cluster')
gene_expression = tidyr::pivot_longer(gene_expression, cols = 1:28, values_to = 'avg_expression', names_to = 'cluster')
plot_data= gene_expression
plot_data$pct_expression= pct_expression_data$pct_expression
cluster_metadata= all_timepoints@meta.data %>% group_by(legend_condition) %>% summarise(cluster= unique(legend_condition), legend= unique(legend), condition= unique(condition) )
plot_data= left_join(plot_data, cluster_metadata[, 2:4])
plot_data= plot_data[plot_data$avg_expression >0.1 & plot_data$gene %in% tnfa_genes,]
plot_data$gene= factor(plot_data$gene, levels = rev(tnfa_genes))
colnames(plot_data)[3:4]= c('Average expression', '% cells expressing')
ggplot(plot_data, aes(y= gene, x= condition, size=`% cells expressing`, col=`Average expression` )) + geom_point() +facet_wrap('legend', nrow = 1) + theme_bw() + RotatedAxis() + scale_color_gradientn(colors = c("orange", "firebrick"))
```
Compute TNFA pathway activity in single cell
```{r eval=FALSE, include=FALSE}
# Rank gene by their expression in each single cell
cells_rankings = AUCell_buildRankings(all_timepoints@assays$SCT@counts, nCores = 40)
# Compute the enrichment of TNFA genes
cells_AUC = AUCell_calcAUC(list(TNFA=tnfa_genes), cells_rankings, aucMaxRank=nrow(cells_rankings)*0.05, nCores = 40)
# Add TNFa score to metadata
all_timepoints$tnfa_score = cells_AUC@assays@data$AUC['TNFA',]
```
## Figure 4.b
```{r}
FeaturePlot(all_timepoints, 'tnfa_score', split.by = 'condition', cols = c('dodgerblue', 'firebrick'), order = T, max.cutoff = 0.15, ncol = 4)
```
## Extended figure 6.b
```{r}
FeaturePlot(all_timepoints, 'tnfa_score', split.by = 'line_condition', cols = c('dodgerblue', 'firebrick'), order = T, max.cutoff = 0.15, ncol = 4)* NoAxes()
```
Importing NFkB direct targets
```{r}
tnfa_direct_targets = read.csv('../gene_lists/nfkb_targets.csv')
tnfa_direct_targets$Human.Gene.Name = gsub("BLIMP1 /PRDM1", "PRDM1", tnfa_direct_targets$Human.Gene.Name)
tnfa_direct_targets$Human.Gene.Name = gsub("SERPINE1, PAI-1", "SERPINE1", tnfa_direct_targets$Human.Gene.Name)
tnfa_genes= unique(tnfa_direct_targets$Human.Gene.Name[tnfa_direct_targets$Human.Gene.Name %in% rownames(all_timepoints)])
```
## Supplementary figure 3.a
```{r}
gene_expression= AverageExpression(subset(all_timepoints, cells = all_timepoints$barcode[all_timepoints$legend %in% levels(all_timepoints$legend)[c(1,3, 5, 9, 16:20)]]), assays = 'SCT', slot = 'counts', features = tnfa_genes, group.by = 'legend_condition')
gene_expression= as.data.frame(gene_expression$SCT)
gene_expression$gene= rownames(gene_expression)
tnfa_genes= tnfa_genes[tnfa_genes %in% rownames(gene_expression)]
pct_expression_data= data.frame(fake_cluster= 1:length(tnfa_genes), row.names = tnfa_genes )
for (cluster in colnames(gene_expression) ){
cells= all_timepoints$barcode[all_timepoints$legend_condition==cluster]
cluster_expression= as.data.frame(all_timepoints@assays$RNA@counts[tnfa_genes, cells]>0)
pct_expression= data.frame(apply(cluster_expression, MARGIN = 1, FUN = function(x){ round(sum(x)/length(x)*100, digits = 0) } ))
colnames(pct_expression)= cluster
pct_expression_data= cbind(pct_expression_data, pct_expression)
}
remove(cluster, pct_expression)
pct_expression_data=pct_expression_data[, 2:29]
pct_expression_data$gene= rownames(pct_expression_data)
pct_expression_data = tidyr::pivot_longer(pct_expression_data, cols = 1:28, values_to = 'pct_expression', names_to = 'cluster')
gene_expression = tidyr::pivot_longer(gene_expression, cols = 1:28, values_to = 'avg_expression', names_to = 'cluster')
#gene_expression$avg_expression = log10(gene_expression$avg_expression+1)
plot_data= gene_expression
plot_data$pct_expression= pct_expression_data$pct_expression
cluster_metadata= all_timepoints@meta.data %>% group_by(legend_condition) %>% summarise(cluster= unique(legend_condition), legend= unique(legend), condition= unique(condition) )
plot_data= left_join(plot_data, cluster_metadata[, 2:4])
plot_data= plot_data[plot_data$avg_expression >0.1 & plot_data$gene %in% tnfa_genes,]
plot_data$gene= factor(plot_data$gene, levels = rev(tnfa_genes))
colnames(plot_data)[3:4]= c('Average expression', '% cells expressing')
plot_data= left_join(plot_data, tnfa_direct_targets[, c('Human.Gene.Name', 'Group')] , by=c('gene'='Human.Gene.Name'))
ggplot(plot_data, aes(y= gene, x= condition, size=`% cells expressing`, col=`Average expression` )) + geom_point() + facet_wrap('legend', nrow = 1) + theme_bw() + RotatedAxis() + scale_color_gradientn(colors = c("orange", "firebrick"))
```
# Analysis of acyclovir treated control (CTRL ACV) organoids
Read digital gene expression matrices (generated with Spacemake) for CTRL samples and create Seurat objects for each sample keeping barcodes with at least 250 detected genes and genes detected at least 5 cells
```{r eval=FALSE, include=FALSE}
Xmoo1_CTRLACV_1 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_CTRLACV_1.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_CTRLACV_1', assay = 'RNA')
Xmoo1_CTRL_1_exonic$line = 'iPSC line 1'
Xmoo1_CTRL_1_exonic$replicate = 'Rep. 1'
Xmoo1_CTRL_1_exonic$condition = 'CTRL ACV'
Xmoo1_CTRLACV_2 = read.table(file = 'digital_gene_expression/scRNAseq_spacemake/Xmoo1_CTRLACV_2.txt.gz', sep = '', row.names = 1, header = T) %>% CreateSeuratObject(min.cells = 5, min.features = 250, project = 'Xmoo1_CTRLACV_2', assay = 'RNA')
Xmoo1_CTRLACV_2$line = 'iPSC line 1'
Xmoo1_CTRLACV_2$replicate = 'Rep. 2'
Xmoo1_CTRLACV_2$condition = 'CTRL ACV'
```
Create a single integrated Seurat object with all CTRL samples (2 replicates for 2 cell lines) after correcting for sample- and cell line-specific batch effects
```{r eval=FALSE, include=FALSE}
integration.list <- list(Xmoo1_CTRLACV_1,Xmoo1_CTRLACV_2, Xmoo1_CTRL_1_exonic, Xmoo1_CTRL_2_exonic, Gline_CTRL_1_exonic, Gline_CTRL_2_exonic)
for (i in 1:length(integration.list)) {
integration.list[[i]] <- SCTransform(integration.list[[i]], verbose = T,method= 'glmGamPoi', assay = 'RNA', new.assay.name = 'SCT', vst.flavor='v2')
}
integration.features <- SelectIntegrationFeatures(object.list = integration.list, nfeatures = 3000)
integration.list <- PrepSCTIntegration(object.list = integration.list, anchor.features = integration.features, verbose = T)
integration.anchors <- FindIntegrationAnchors(object.list = integration.list, normalization.method = "SCT", anchor.features = integration.features, verbose = T)
CTRL_ACV <- IntegrateData(anchorset = integration.anchors, normalization.method = "SCT", verbose = T)
DefaultAssay(CTRL_ACV) <- 'integrated'
remove(i, integration.list, integration.features, integration.anchors)
```
Perform dimensionality reduction and clustering with 30 principal components and default parameters
```{r eval=FALSE, include=FALSE}
CTRL_ACV <- CTRL_ACV %>% RunPCA( verbose = T) %>% RunUMAP( dims = 1:30, verbose = T) %>% FindNeighbors( dims = 1:30, verbose = T) %>% FindClusters( verbose = T)
```
Inspect transcript counts (nCount) and % of mitochondrial transcripts (percent_MT) to identify low quality clusters
```{r eval=FALSE, include=FALSE}
VlnPlot(CTRL_ACV, 'nCount_RNA', y.max = 2000, pt.size = 0)+ NoLegend()
VlnPlot(CTRL_ACV, 'percent_MT', y.max = 2000, pt.size = 0)+ NoLegend()
```
Remove low quality clusters 2 (low nCount) and 11 (low nCount, high mito) and select top 75% cells by nCount per cluster
```{r eval=FALSE, include=FALSE}
CTRL_ACV$barcode= colnames(CTRL_ACV)
top75= CTRL_ACV@meta.data[!CTRL_ACV$seurat_clusters %in% c(2, 11),] %>% group_by(seurat_clusters) %>% top_frac(n = 0.75, wt = nCount_RNA_exonic)
CTRL_ACV = subset(CTRL_ACV, cells=top75$barcode)
```
Re-run integration, dimensionality reduction and clustering
```{r eval=FALSE, include=FALSE}
Idents(CTRL_ACV) = 'sample'
integration.list <- SplitObject(CTRL_ACV)
for (i in 1:length(integration.list)) {
integration.list[[i]] <- SCTransform(integration.list[[i]],method= 'glmGamPoi', assay = 'RNA', new.assay.name = 'SCT', vst.flavor='v2')
}
integration.features <- SelectIntegrationFeatures(object.list = integration.list, nfeatures = 3000)
integration.list <- PrepSCTIntegration(object.list = integration.list, anchor.features = integration.features)
integration.anchors <- FindIntegrationAnchors(object.list = integration.list, normalization.method = "SCT", anchor.features = integration.features)
CTRL_ACV <- IntegrateData(anchorset = integration.anchors, normalization.method = "SCT")
DefaultAssay(CTRL_ACV) <- 'integrated'
CTRL_ACV <- CTRL_ACV %>% RunPCA() %>% RunUMAP( dims = 1:30) %>% FindNeighbors( dims = 1:30) %>% FindClusters()
remove(i, integration.list, integration.features, integration.anchors)
```
Add annotations of CTRL cells
```{r eval=FALSE, include=FALSE}
CTRL_ACV$legend_CTRL = CTRL$legend
```
Annotate clusters accordingly
```{r eval=FALSE, include=FALSE}
Idents(CTRL_ACV) <- 'seurat_clusters'
CTRL_ACV <- RenameIdents(
object = CTRL_ACV,
'0'= 'Immature Cortical Neurons 2',
'1'= 'Radial Glia G1/S Phase',
'2'= 'Immature Cortical Neurons 1',
'3'= 'Radial Glia G2M Phase',
'4'= 'Astroglia',
'5'= 'Mature Cortical Neurons',
'6'= 'Intermediate Progenitors',
'7'= 'Hindbrain Neurons 1',
'8'= 'Radial Glia G1/S Phase',
'9'= 'Cortical Hem/Choroid Plexus',
'10'= 'Thalamic Neurons',
'11'= 'Intermediate Progenitors',
'12'= 'Mature Cortical Neurons',
'13'= 'Retinal Progenitors',
'14'= 'Thalamic Neurons',
'15'= 'Astroglia',
'16'= 'Retinal Pigmented Cells',
'17'= 'Progenitors',
'18'= 'Progenitors',
'19'= 'Mural Cells',
'20'= 'Cortical Hem/Choroid Plexus',
'21'= 'Hindbrain Neurons 2',
'22'= 'Hindbrain Neurons 2',
'23'= 'Cajal Retzius Neurons')
CTRL_ACV$legend = Idents(CTRL_ACV)
```
Import processed object
```{r}
CTRL_ACV= readRDS('seurat_objects/CTRL_ACV.rds')
```
## Extended figure 3.b
```{r}
DimPlot(CTRL_ACV, order = T, split.by = 'condition', cols = color_palette)
```
Compute TNFA pathway activity in single cells
```{r eval=FALSE, include=FALSE}
# Rank gene by their expression in each single cell
cells_rankings = AUCell_buildRankings(CTRL_ACV@assays$SCT@counts, nCores = 40)
# Compute the enrichment of TNFA genes
cells_AUC = AUCell_calcAUC(list(TNFA=tnfa_genes), cells_rankings, aucMaxRank=nrow(cells_rankings)*0.05, nCores = 40)
# Add TNFa score to metadata
CTRL_ACV$tnfa_score = cells_AUC@assays@data$AUC['TNFA',]
remove(cells_rankings, cells_AUC)
```
## Extended figure 6.a
```{r}
rbind(all_timepoints@meta.data[, c('tnfa_score', 'condition')], CTRL_ACV@meta.data[CTRL_ACV$condition== 'CTRL ACV', c('tnfa_score', 'condition')] ) %>%
ggplot(aes(x=condition, y=tnfa_score, fill= condition)) + geom_boxplot() + ylim(c(0, 0.15)) +
theme_bw()
```
# Analysis of bulk RNAseq data from CTRL and HSV1 3 day post infection (3dpi) organoids
Import gene expression counts and sample metadata after alignment with PigX pipeline
```{r}
countData= read.csv('digital_gene_expression/bulkRNAseq_pigx/feature_counts.tsv', sep = '\t')
colData= read.csv('digital_gene_expression/bulkRNAseq_pigx/colData.tsv', sep = '\t')[colnames(countData), ]
```
Run differential gene expression using DESeq2
```{r}
# Prepare DESeq 2 object and exclude cell line specific effects
bulk_data <- DESeqDataSetFromMatrix(countData = countData,
colData = colData,
design = ~ line + group)
# Filter lowly expressed genes
bulk_data <- bulk_data[rowSums(counts(bulk_data)) >= 10,]
# Fit model
bulk_data <- DESeq(bulk_data)
# Extract logFC and expression for all genes
differential_results <- results(bulk_data)
differential_results= data.frame(differential_results@listData, gene = rownames(bulk_data), row.names = rownames(bulk_data))
remove(countData, colData, bulk_data)
```
Identify synaptic genes
```{r}
synaptic.dt = gconvert(query = "GO:0007268")
synaptic.dt= synaptic.dt[synaptic.dt$target %in% rownames(countData),]
```
## Extended figure 2.b
```{r}
syn.labels <- c("ARC", "NPTX2", "EGR2", "SV2B", "HTR2A",
"GRIM3", "SV2A", "SCN2B", "ADCY1", "CALB2",
"CPLX2", "SYT3", "MAP1A", "MAPT", "GABBR2",
"SEPT5", "NLGN3", "RIMS1", "RIMS4", "SST",
"SYP", "VAMP2", "GRIN1", "CHARM3", "SNAP25")
synaptic.selected = differential_results[synaptic.dt$target[synaptic.dt$name %in%syn.labels],]
synaptic.selected$symbol = synaptic.dt$name[synaptic.dt$name %in%syn.labels]
ggplot(differential_results, aes(x= log10(baseMean+1), y= log2FoldChange)) +
geom_point(size=0.01, col= 'lightgrey') +
geom_point(data= differential_results[grepl(pattern = 'HSV', differential_results$gene),],
mapping= aes(x= log10(baseMean+1), y= log2FoldChange),size= 0.5,col= 'darkgreen')+
#geom_density2d(data = differential_results[synaptic.dt$target, ],
# mapping = aes(x= log10(baseMean+1), y= log2FoldChange), col='black') +
geom_hline(yintercept = 0, linetype=2) +
geom_point(data = differential_results[synaptic.dt$target, ],
mapping = aes(x= log10(baseMean+1), y= log2FoldChange), size= 0.5, col= 'red') +
theme(panel.grid.major = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_blank())+
ggtitle('HSV1 3dpi vs CTRL (BulkRNAseq)')
remove(synaptic.dt, synaptic.selected)
```
Identify TNF ligands
```{r}
# Import TNF ligands from HGNC database
tnf_ligands= read.table('gene_lists/tnf_ligands_HGNC.csv', header = T, sep=',')[, 'Approved.symbol']
#Convert gene symbols to ensemble ID
library("org.Hs.eg.db")
tnf_ligands_ensembleID <- mapIds(org.Hs.eg.db, keys = tnf_ligands, keytype = "SYMBOL", column="ENSEMBL")
# Add gene symbols for TNF ligands
differential_results$gene_symbol= NA
differential_results[tnf_ligands_ensembleID , 'gene_symbol' ]= names(tnf_ligands_ensembleID)
```
## Extended figure 7.a
```{r}
# Plot FC against mean expression for all genes and highlight viral genes and TNF ligands
ggplot(differential_results, aes(x= log10(baseMean+1), y= log2FoldChange)) +
geom_point(size= 0.01, col= 'lightgrey') +
geom_point(data= differential_results[grepl(pattern = 'HSV', differential_results$gene),],
mapping= aes(x= log10(baseMean+1), y= log2FoldChange),size= 1,col= 'darkgreen')+
geom_point(data= differential_results[differential_results$gene %in% tnf_ligands_ensembleID,],
mapping= aes(x= log10(baseMean+1), y= log2FoldChange),size= 1,col= 'firebrick')+
geom_hline(yintercept = 0, linetype=2) +
ggrepel::geom_text_repel(data= differential_results[differential_results$log2FoldChange>2.5, ],
mapping = aes(x= log10(baseMean+1), y= log2FoldChange, label= gene_symbol), min.segment.length = 0, max.overlaps = 30)+
theme(panel.grid.major = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_blank())+
ggtitle('HSV1 3dpi vs CTRL (BulkRNAseq)')
remove(tnf_ligands,tnf_ligands_ensembleID)
```
## Extended figure 7.b
```{r}
# Extract sample level pseudobulk profiles from single cell data
sc_pseudobulk= t(rowsum(x = t(as.matrix(all_timepoints@assays$RNA@counts)), group = paste0(all_timepoints$condition, '_', all_timepoints$line, '_', all_timepoints$replicate) ))
# Define sample level metadata
sc_colData= data.frame(group= c('CTRL','CTRL','CTRL','CTRL', 'inf1dpi','inf1dpi','inf1dpi','inf1dpi', 'inf3dpi', 'inf3dpi','inf3dpi', 'inf3dpi', 'inf3dpi_ACV','inf3dpi_ACV', 'inf3dpi_ACV','inf3dpi_ACV'), line= c('line 1','line 1','line 2','line 2', 'line 1','line 1','line 2','line 2','line 1','line 1','line 2','line 2','line 1','line 1','line 2','line 2'), row.names = colnames(sc_pseudobulk) )
# Define upregulated TNF ligands
upregulated_tnf= names(tnf_ligands_ensembleID)[tnf_ligands_ensembleID %in% differential_results$gene[differential_results$log2FoldChange>2.5]]
# Plot heatmap of upregulated tnf ligands detected in single cell data
pheatmap(sc_pseudobulk[rownames(sc_pseudobulk) %in% upregulated_tnf, rownames(sc_colData) ], annotation_col = sc_colData, cluster_cols = F, show_colnames = F, scale='row',
color= brewer.pal(9,'Blues'),
annotation_colors = list(group= c(CTRL= 'grey' , inf1dpi= 'lightgreen', inf3dpi= 'darkgreen' , inf3dpi_ACV='orange' )),
height = 7, width = 7, main = 'Upreglated TNF ligands detected in scRNAseq (pseudobulk)')
remove(sc_pseudobulk, sc_colData, upregulated_tnf, differential_results)
```
点赞本文的读者
还没有人对此文章表态
没有评论
RNA-seq data analysis (R-part) for S. epidermidis 1585
RNA-seq recalculate p-values on sT- and LT-transcripts
© 2023 XGenes.com Impressum