tme-deg-gsea.Rmd

# **Tumor ecosystem analysis**

## Loading packages

```{r, eval=TRUE, warning=FALSE, message=FALSE}
library(IOBR)
```

## Downloading data for example
Obtaining data set from GEO [Gastric cancer: GSE62254](https://pubmed.ncbi.nlm.nih.gov/25894828/) using `GEOquery` R package.
```{r,message=FALSE,warning=FALSE}
if (!requireNamespace("GEOquery", quietly = TRUE))  BiocManager::install("GEOquery")
library("GEOquery")
# NOTE: This process may take a few minutes which depends on the internet connection speed. Please wait for its completion.
eset_geo<-getGEO(GEO     = "GSE62254", getGPL  = F, destdir = "./")
eset    <-eset_geo[[1]]
eset    <-exprs(eset)
eset[1:5,1:5]
```

## Gene Annotation: HGU133PLUS-2 (Affaymetrix)
```{r,message=FALSE,warning=FALSE}

# Conduct gene annotation using `anno_hug133plus2` file; If identical gene symbols exists, these genes would be ordered by the mean expression levels. The gene symbol with highest mean expression level is selected and remove others. 

eset<-anno_eset(eset       = eset,
                annotation = anno_hug133plus2,
                symbol     = "symbol",
                probe      = "probe_id",
                method     = "mean")
eset[1:5, 1:3]
```

## Determine TME subtype of gastric cancer using [TMEclassifier R package](https://github.com/LiaoWJLab/TMEclassifier)
```{r, message=FALSE, warning=FALSE}
if (!requireNamespace("TMEclassifier", quietly = TRUE))  devtools::install_github("LiaoWJLab/TMEclassifier")
library(TMEclassifier)
tme <- tme_classifier(eset = eset, scale = TRUE)
table(tme$TMEcluster)
head(tme)
```

```{r}
table(tme$TMEcluster)
head(tme)
```

## DEG analysis: method1
Differential analysis of selected immune-activated and immune-expelled gastric cancers

```{r}
pdata <- tme[!tme$TMEcluster=="IS", ]
deg  <-   iobr_deg(eset         = eset,
                   annotation    = NULL,
                   pdata        = pdata,
                   group_id     = "TMEcluster",
                   pdata_id     = "ID",
                   array        = TRUE,
                   method       = "limma",
                   contrast     = c("IA","IE"),
                   path         = "result",
                   padj_cutoff  = 0.01,
                   logfc_cutoff = 0.5)
```

## GSEA analysis based on differential express gene analysis results


Select the gene set list in IOBR's signature collection.
```{r}
head(deg)

sig_list <- signature_collection[c("TMEscoreB_CIR", "TMEscoreA_CIR", "DNA_replication", "Base_excision_repair",
                                   "Pan_F_TBRs", "TGFb.myCAF", "Ferroptosis", "TLS_Nature", "Glycolysis")]
sig_list
```

```{r, eval=FALSE}
gsea<-     sig_gsea(deg,
                    genesets          = sig_list,
                    path              = "result",
                    gene_symbol       = "symbol",
                    logfc             = "log2FoldChange",
                    org               = "hsa",
                    show_plot         = FALSE,
                    msigdb            = TRUE,
                    category          = "H",
                    subcategory       = NULL,
                    palette_bar       = "set2")
```

```{r echo=FALSE, fig.cap='GSEA of TME gent sets', out.width='95%', fig.asp=.85, fig.align='center'}
knitr::include_graphics(rep("./fig/gsea-1.png", 1))
```

Hallmark gene signatures
```{r, eval=FALSE}
gsea<-     sig_gsea(deg,
                    genesets          = NULL,
                    path              = "GSEA",
                    gene_symbol       = "symbol",
                    logfc             = "log2FoldChange",
                    org               = "hsa",
                    show_plot         = FALSE,
                    msigdb            = TRUE,
                    category          = "H",
                    subcategory       = NULL,
                    palette_bar       = "aaas",
                    show_bar          = 5,
                    show_gsea         = 6)
```

```{r echo=FALSE, fig.cap='GSEA of Hallmark gent sets', out.width='95%', fig.asp=.85, fig.align='center'}
knitr::include_graphics(rep("./fig/gsea-2.png", 1))
```


## DEG analysis: method2

Identifing TME subtype-related differential genes using `find_markers_in_bulk`
```{r, message = F, warning = F}
library(Seurat)
res <- find_markers_in_bulk(pdata      = tme, 
                            eset       = eset, 
                            group      = "TMEcluster", 
                            nfeatures  = 2000, 
                            top_n      = 50, 
                            thresh.use = 0.15, 
                            only.pos   = TRUE, 
                            min.pct    = 0.10)
top15 <-  res$top_markers %>% dplyr:: group_by(cluster) %>%  dplyr::top_n(15, avg_log2FC)
top15$gene
```

Heatmap visualisation using `Seurat`'s `DoHeatmap`
```{r,fig.show= 'hide',message=FALSE}
#定义分型对应的颜色
cols <- c('#2692a4','#fc0d3a','#ffbe0b')
p1 <- DoHeatmap(res$sce, top15$gene, group.colors = cols )+
  scale_fill_gradientn(colours = rev(colorRampPalette(RColorBrewer::brewer.pal(11,"RdBu"))(256)))

```

Extracting variables from the expression matrix to merge with TME subtypes

```{r,fig.show= 'hide',message=FALSE}

input <- combine_pd_eset(eset = eset, pdata = tme, feas = top15$gene, scale = T)
p2 <- sig_box(input, variable = "TMEcluster", signature = "IFNG", jitter = TRUE,
              cols =  cols, show_pvalue = TRUE, size_of_pvalue = 4)

p3 <- sig_box(input, variable = "TMEcluster", signature = "IL1A", 
              jitter = TRUE, cols =  cols, show_pvalue = TRUE, size_of_pvalue = 4)
```

```{r, fig.width= 12, fig.height=9, message = F, warning = F}

if (!requireNamespace("patchwork", quietly = TRUE))   install.packages("patchwork")
library(patchwork)
p <- (p1|p2/p3) + plot_layout(widths = c(2.3,1))
p + plot_annotation(tag_levels = 'A')
```


## Identifying signatures associated with TME clusters

Calculate TME associated signatures-(through PCA method).
```{r, message = F, warning = F}
sig_tme<-calculate_sig_score(pdata           = NULL,
                             eset            = eset,
                             signature       = signature_collection,
                             method          = "pca",
                             mini_gene_count = 2)
sig_tme <- t(column_to_rownames(sig_tme, var = "ID"))
sig_tme[1:5, 1:3]
```

Finding signatures or cell types associated with TMEcluster
```{r, message = F, warning = F}
res <- find_markers_in_bulk(pdata = tme, eset = sig_tme, group = "TMEcluster", nfeatures = 1000, top_n = 20, min.pct = 0.10)

top15 <-  res$top_markers %>% dplyr:: group_by(cluster) %>%  dplyr::top_n(15, avg_log2FC)

p1 <- DoHeatmap(res$sce, top15$gene, group.colors = cols)+
  scale_fill_gradientn(colours = rev(colorRampPalette(RColorBrewer::brewer.pal(11,"RdBu"))(256)))
```

```{r,fig.show= 'hide', message=FALSE}
top15$gene  <- gsub(top15$gene, pattern = "-", replacement = "\\_")
input <- combine_pd_eset(eset = sig_tme, pdata = tme, feas = top15$gene, scale = T)

p2 <- sig_box(input, variable = "TMEcluster", signature = "CD_8_T_effector", jitter = TRUE,
              cols =  cols, show_pvalue = TRUE, size_of_pvalue = 4, size_of_font = 6)

p3 <- sig_box(input, variable = "TMEcluster", signature = "Neutrophils_Bindea_et_al",  
              jitter = TRUE, cols =  cols, show_pvalue = TRUE, size_of_pvalue = 4, size_of_font = 6)
```


```{r, fig.width= 12, fig.height=9}
p <- (p1|p2/p3) + plot_layout(widths = c(2.3,1))
p + plot_annotation(tag_levels = 'A')
```


```{r, fig.width= 5, fig.height=8}
library(survminer)
data(pdata_acrg, package = "IOBR")
input <- merge(pdata_acrg, input, by = "ID")
p1<-surv_group(input_pdata       = input,
               target_group      = "TMEcluster",
               ID                = "ID",
               reference_group   = "High",
               project           = "ACRG",
               cols              = cols, 
               time              = "OS_time",
               status            = "OS_status",
               time_type         = "month",
               save_path         = "result")
p1
```

```{r,fig.show= 'hide',message=FALSE}
p1<- percent_bar_plot(input, x = "TMEcluster" , y = "Subtype", palette = "jama")
p2<- percent_bar_plot(input, x = "TMEcluster" , y = "Lauren", palette = "jama")
p3<- percent_bar_plot(input, x = "TMEcluster" , y = "TMEscore_binary", palette = "jama")
```

```{r,fig.width= 12, fig.height=8}
p1|p2|p3
```

## References

Cristescu, R., Lee, J., Nebozhyn, M. et al. Molecular analysis of gastric cancer identifies subtypes associated with distinct clinical outcomes. Nat Med 21, 449–456 (2015). https://doi.org/10.1038/nm.3850

[CIBERSORT](https://cibersort.stanford.edu/); Newman, A. M., Liu, C. L., Green, M. R., Gentles, A. J., Feng, W., Xu, Y., … Alizadeh, A. A. (2015). Robust enumeration of cell subsets from tissue expression profiles. Nature Methods, 12(5), 453–457.  https://doi.org/10.1038/nmeth.3337; 

Seurat: Hao and Hao et al. Integrated analysis of multimodal single-cell data. Cell (2021)

Zeng D, Yu Y, Qiu W, Mao Q, …, Zhang K, Liao W; Tumor microenvironment immunotyping heterogeneity reveals distinct molecular mechanisms to clinical immunotherapy applications in gastric cancer. (2024) Under Review.