analysis_markdown_diagnosis_comparison.Rmd

---
title: "nCounter_RNA_Analysis"
author: "Ned Cauley"
date: "8/5/24"
output: html_document
---

# This markdown and analysis based on https://github.com/bhattacharya-a-bt/CBCS_normalization/blob/master/sabry_analysis.R
# The above analysis is from the publication https://academic.oup.com/bib/article/22/3/bbaa163/5891144?login=true

```{r setup}
knitr::opts_chunk$set(echo = TRUE)
```

## Set up libraries and QC functions libraries

```{r Libraries and RUVseq functions, message=FALSE, warning=FALSE}
#library(NanoStringQCPro)
#library(NanoStringNorm)

library(ggplot2)
library(EnvStats)
library(MASS)
library(RUVSeq)
library(DESeq2)
library(limma)
library(matrixStats)
library(readxl)
library(plotly)
library(pheatmap)
library(ggrepel)
library(gridExtra)
library(grid)
library(dplyr)
library(tibble)
library(tidyr)
library(stringr)
library(ggfortify)
library(PCAtools)
library(ggpubr)

source("R/ncounter_analysis_functions.R")

# Load the annotation file
annotation_file_path <- "5_20_24_ncounter_rna_annotation_edit_NC.csv"
annotation.df <- read.csv(annotation_file_path)


# Create a new folder for the run date and results
run.folder <- "11_13_2024/diagnosis_comparison/"

results.folder <- "results/"

```

```{r Define the contrast for the DEG analysis, message=FALSE, warning=FALSE}

# Define the contrast and levels from the annotation columns for QC and DEG analysis
# Levels are: Effect, Reference. 
# This means a positive log2foldchange = lower reference, higher effect
comparison.list <- list(
  "Diagnosis_MCD" = c("KS_MCD", "KS"), 
  "Diagnosis_PEL" = c("KS_PEL", "KS"), 
  "Diagnosis_KICS" = c("KS_KICS", "KS"), 
  "Diagnosis_MCD_PEL" = c("KS_MCD_PEL", "KS"))

```


## Load the processed read counts

```{r Initial Dataframes and QC,  message=FALSE, warning=FALSE}

load(file = "ncounter_processed_rcc_counts.rda")

```

## Run RUVseq differential expression analysis

```{r, message=FALSE, warning=FALSE}

# Toggles for analysis and plots
run.de <- TRUE
export.de.results <- FALSE
export.norm.counts <- FALSE
create.volcano <- TRUE
export.volcano <- FALSE
create.heatmap <- TRUE
export.heatmap <- FALSE
export.PCA <- TRUE


# Start a vector of HK genes to use
hk.final.list <- c("UBB", 
                   "TLK2", 
                   "DNAJC14", 
                   "POLR2A", 
                   "SDHA", 
                   "STK11IP", 
                   "OAZ1", 
                   "TBC1D10B", 
                   "GUSB", 
                   "PSMC4", 
                   "TFRC", 
                   "ABCF1", 
                   "G6PD")

# Retrieve KSHV genes
kshv.genes <- rownames(rcc.dfs$raw.counts[grep("KSHV", rownames(rcc.dfs$raw.counts)), ])
kshv.genes <- c(kshv.genes, "ORF75")

# Toggle to rerun analysis
de.results.list <- list()

# Run each DE contrast
if(run.de == TRUE){
  
  for(comparison in names(comparison.list)){
    
    contrast.levels <- comparison.list[[comparison]]
    
    # Run DESEQ with RUVseq normalization for k sources of variability
    deseq.output <- run_RUV_DESEQ2(contrast = "Diagnosis", 
                   contrast.levels = contrast.levels, 
                   annotation.data = annotation.df, 
                   raw.counts = rcc.dfs$raw.counts, 
                   feature.data = rcc.dfs$fData, 
                   k = 1, 
                   additional.covar = NULL, 
                   hk.genes = hk.final.list)
    
    de.results.list[[comparison]] <- deseq.output
    
    
    # Export all DE results
    if(export.de.results == TRUE){
      write.csv(deseq.output$all.de.results, file = paste0(results.folder, run.folder,   "DE_results_",comparison,".csv"), row.names =   FALSE)
    }
    
    # Gather normalized counts
    normalized.counts <- as.data.frame(counts(de.results.list[[comparison]]$dds, normalized = TRUE))
    # Add a gene column
    gene <- rownames(normalized.counts)
    normalized.counts <- cbind(gene, normalized.counts) 
  
  }
  
  # Export Normalized read counts
  if(export.norm.counts == TRUE){

    write.csv(normalized.counts, file = paste0(results.folder, 
    run.folder,   
    "RUV_normalized_counts.csv"), 
    row.names = FALSE)
    
  }

}
```

### Plots

#### Volcano Plot

```{r}
# Create plots per contrast

# Label cytokine genes
cytokine.genes <- normalized.counts[grep("^IL", normalized.counts$gene), ]$gene

cytokine.genes <- c(cytokine.genes, "TNF", "IFNA1", "IFNG", "IRF4")

# Volcano plot
if(create.volcano == TRUE){
  
  volcano.list <- list()
  custom.volcano.list <- list()
  
  for(comparison in names(de.results.list)){
    
    contrast.levels <- comparison.list[[comparison]]
    
    # Standard volcano
    volcano.output <- make_volcano(de.results = de.results.list[[comparison]]$all.de.results,
                                   title = gsub("_", " ", comparison), 
                                   legend.title = paste0("Expression in ", contrast.levels[1]), 
                                   x.axis.title = "log2 Fold Change", 
                                   fc.limit = 1, 
                                   custom.gene.labels = NULL, 
                                   remove.controls = TRUE, 
                                   remove.genes = kshv.genes)
    
    volcano.list[[comparison]] <- volcano.output
    
    # Export the volcano plot
    if(export.volcano == TRUE){
        ggsave(paste0(results.folder, run.folder, "volcano_", comparison, "_adjpval.png"),
             plot = volcano.output$volcano.plot, 
             width = 8, 
             height = 8, 
             units = "in")
    }
    
    # Custom volcano with cytokine genes labeled
    volcano.output <- make_volcano(de.results = de.results.list[[comparison]]$all.de.results,
                                   title = gsub("_", " ", comparison), 
                                   legend.title = paste0("Expression in ", contrast.levels[1]), 
                                   x.axis.title = "log2 Fold Change", 
                                   fc.limit = 1, 
                                   custom.gene.labels = cytokine.genes, 
                                   remove.controls = TRUE, 
                                   remove.genes = kshv.genes)
    
    custom.volcano.list[[comparison]] <- volcano.output
    
    # Export the volcano plot
    if(export.volcano == TRUE){
        ggsave(paste0(results.folder, run.folder, "cytokines_volcano_", comparison, "_adjpval.png"),
             plot = volcano.output$volcano.plot, 
             width = 8, 
             height = 8, 
             units = "in")
    }
  
    
  }
  
}

```

#### Heatmap

```{r}

# Heatmap
if(create.heatmap == TRUE){
  
  heatmap.list <- list()
  
  for(comparison in names(de.results.list)){
    
    contrast.levels <- comparison.list[[comparison]]
  
    heatmap.output <- make_heatmap(deseq.output = de.results.list[[comparison]], 
                                    contrast = "Diagnosis", 
                                    contrast.levels = contrast.levels, 
                                    annotation = annotation.df, 
                                    remove.genes = kshv.genes, 
                                    cluster.rows = TRUE, 
                                    show.row.names = TRUE, 
                                    cluster.cols = TRUE, 
                                    show.col.names = FALSE)
    
    heatmap.list[[comparison]] <- heatmap.output
    
    
    if(export.heatmap == TRUE){
    
      ggsave(paste0(results.folder, run.folder, "heatmap_", comparison, ".png"),
           plot = heatmap.output$heatmap.contrast, 
           width = 12, 
           height = 12, 
           units = "in")
      }
    
    
    
  }
  
}


```

# Load previously generated normalized counts

```{r}

normalized.counts <- read.csv(paste0(results.folder, 
                                     "11_13_2024/all/", 
                                     "RUV_normalized_counts.csv"))

# Fix column names to match annotation sheet
colnames(normalized.counts) <- gsub("\\."," ", colnames(normalized.counts))

# Add row names
rownames(normalized.counts) <- normalized.counts$gene

```

# Read count information for specific genes

```{r}

#Create a combined violin and boxplot

# Cytokine genes
cytokine.genes <- normalized.counts[grep("^IL", normalized.counts$gene), ]$gene

cytokine.genes <- c(cytokine.genes, "TNF", "IFNA1", "IFNG", "IRF4")

KS.alone.cytokine.boxplot <- gene_counts_violin_boxplot(counts = normalized.counts,
                                               annotation.df = annotation.df, 
                                               gene.list = cytokine.genes,
                                               annotation.field = "KS_alone", 
                                               display.summary.stat = FALSE, 
                                               compare.groups = TRUE)

multi.diagnosis.groups <- c("KS_KICS", 
                           "KS_MCD", 
                           "KS_MCD_PEL", 
                           "KS_PEL")

cytokine.boxplot.list <- list()

for(comparison in multi.diagnosis.groups){
  
  compare.groups <- c("KS", comparison)
  
  annotation.subset <- annotation.df %>% 
    filter(Diagnosis %in% compare.groups)
  
  group.boxplot <- gene_counts_violin_boxplot(counts = normalized.counts,
                                               annotation.df = annotation.subset, 
                                               gene.list = cytokine.genes,
                                               annotation.field = "Diagnosis", 
                                               display.summary.stat = FALSE, 
                                               compare.groups = TRUE)
  
  cytokine.boxplot.list[[group]] <- group.boxplot
  
  export.cytokine.boxplot <- TRUE
  if(export.cytokine.boxplot == TRUE){
    
    ggsave(paste0(results.folder, run.folder, "cytokine_boxplot_", comparison, ".png"),
           plot = group.boxplot, 
           width = 12, 
           height = 12, 
           units = "in")
    
  }
  
}




```

# PCA

```{r}

# Plot PCA

# Run for each diagnosis comparison

PCA.plot.list <- list()

for(comparison in names(comparison.list)){
  
  contrast.levels <- comparison.list[[comparison]]
  
  # Filter annotation for specific contrast levels
  subset.annotation.data <- annotation.df %>% 
    filter(Diagnosis %in% contrast.levels)
  
  annotation.samples <- subset.annotation.data$SampleID
  
  rownames(subset.annotation.data) <- annotation.samples
  
  # Log transform the counts
  rlog.dds <- rlog(de.results.list[[comparison]]$dds)
  
  # PCA Plot
  PCA.plot <- plotPCA(rlog.dds, intgroup = "Diagnosis", ntop = 700)
  
  if(export.PCA == TRUE){
    
    ggsave(paste0(results.folder, run.folder, comparison, "_PCA.pdf"), 
           plot = PCA.plot)
    
  }
  
  PCA.plot.list[[comparison]] <- PCA.plot
  
}

# Run PCA for cytokine genes

# cytokine gene only counts
cytokine.counts <- normalized.counts %>% 
  filter(gene %in% cytokine.genes)

cytokine.PCA.plot.list <- list()

for(comparison in names(comparison.list)){
  
  contrast.levels <- comparison.list[[comparison]]
  
  # Filter annotation for specific contrast levels
  subset.annotation.data <- annotation.df %>% 
    filter(Diagnosis %in% contrast.levels)
  
  annotation.samples <- subset.annotation.data$SampleID
  
  rownames(subset.annotation.data) <- annotation.samples
  
  # Filter the raw counts for the same samples
  subset.cytokine.counts <- cytokine.counts %>% 
    select(all_of(annotation.samples))
  
  # Log transform the counts
  log2.counts <- subset.cytokine.counts %>% 
    mutate(across(where(is.numeric), log2))
  
  # PCA table
  PCA.table <- pca(log2.counts, 
                 metadata = subset.annotation.data, 
                 removeVar = 0.1)
  
  PCA.plot <- biplot(PCA.table, 
                     colby = "Diagnosis", 
                     legendPosition = "right", 
                     legendLabSize = 10, 
                     legendIconSize = 5, 
                     lab = NULL,
                     title = paste0(gsub("_", " ", comparison), " vs. KS"), 
                     subtitle = "Cytokine Genes Only")
  
  if(export.PCA == TRUE){
    
    ggsave(paste0(results.folder, run.folder, comparison, "_cytokines_PCA.pdf"), 
           plot = PCA.plot)
    
  }
  
  cytokine.PCA.plot.list[[comparison]] <- PCA.plot
  
}


```