mixture.Rmd

---
title: 'DGE Analysis Notebook: BL6 Left vs Right'
output:
  html_document: 
    toc: yes
    fig_width: 8
    fig_height: 8
    fig_caption: yes
    number_sections: yes
    toc_depth: 4
    df_print: tibble
---

```{r, message = FALSE, warning = FALSE}
suppressPackageStartupMessages(library("dplyr"))
suppressPackageStartupMessages(library("DESeq2"))
suppressPackageStartupMessages(library("pheatmap"))
suppressPackageStartupMessages(library("PoiClaClu"))
suppressPackageStartupMessages(library("RColorBrewer"))
suppressPackageStartupMessages(library('tidyverse'))
suppressPackageStartupMessages(library("PoiClaClu"))
suppressPackageStartupMessages(library("vsn"))
suppressPackageStartupMessages(library('EnhancedVolcano'))
suppressPackageStartupMessages(library('gplots'))
suppressPackageStartupMessages(library('org.Mm.eg.db'))
suppressPackageStartupMessages(library('stringr'))
suppressPackageStartupMessages(library("genefilter"))
suppressPackageStartupMessages(library("dplyr"))
suppressPackageStartupMessages(library("ggplot2"))
suppressPackageStartupMessages(library("glmpca"))
suppressPackageStartupMessages(library('org.Mm.eg.db'))
suppressPackageStartupMessages(library("AnnotationDbi"))
suppressPackageStartupMessages(library("apeglm"))
suppressPackageStartupMessages(library("ComplexHeatmap"))
suppressPackageStartupMessages(library("clusterProfiler"))
suppressPackageStartupMessages(library('ggrepel'))
suppressPackageStartupMessages(library('corrplot'))
suppressPackageStartupMessages(library("GO.db"))
suppressPackageStartupMessages(library('edgeR'))
suppressPackageStartupMessages(library('GOstats'))
suppressPackageStartupMessages(library('pathview'))
suppressPackageStartupMessages(library("gage"))
suppressPackageStartupMessages(library("gageData"))
suppressPackageStartupMessages(library('GOSemSim'))
suppressPackageStartupMessages(library('DOSE'))
suppressPackageStartupMessages(library('enrichplot'))
suppressPackageStartupMessages(library('ggnewscale'))
suppressPackageStartupMessages(library('glue'))
suppressPackageStartupMessages(library("ggupset"))
#suppressPackageStartupMessages(library("SPIA"))
suppressPackageStartupMessages(library("stats"))
suppressPackageStartupMessages(library("FactoMineR"))
suppressPackageStartupMessages(library("factoextra"))
suppressPackageStartupMessages(library("pcaExplorer"))
```

# Differential Gene Expression Analysis

## Creating metadata for the DGE Analysis

DESeq2 needs sample information (metadata) for performing DGE analysis.
Let's create the sample information

```{r}
# Read the csv file and change the column name. the samples.csv is a list of sample names, ie, the names of bam files.
sample_ID <- read.csv("~/R/Alina_RNAseq/mixturesamples.csv")
head(sample_ID)
```

```{r}
condition <- c("Infected", "Infected", "control", "control",
               "Infected", "Infected", "control", "control",
               "Infected", "Infected", "control", "control"  )
coldata <- data.frame(sample_ID, condition)
colnames(coldata) <- c('Sample_Name','condition') # change name of one of the columns
# The metadata can be found in a df called coldata!
head(coldata)
```

### Tidying up the names for plots later!

#### First from coldata

```{r}
# tidying up the names od samples in both columns that list of samples
#coldata$Samples <- str_remove_all(coldata$Sampls, pattern = "run6_trimmed_|_.bam|_S\\d\\d|_S\\d")
# coldata$Sample_Name <- str_remove_all(coldata$Sample_Name,
#                                   pattern = "run6_trimmed_|_.bam|_S\\d\\d|_S\\d" |  )
coldata$condition <- as.factor(coldata$condition)
```

Changing the names of samples (as per Alina)

```{r}
# coldata[coldata == '476_R1'] <- 'T'
# coldata[coldata == '754_R1'] <- 'S54'
# coldata[coldata == '755_R1'] <- 'S55'
# coldata[coldata == '757_R1'] <- 'L57'
# coldata[coldata == '758_R1'] <- 'A58'
# coldata[coldata == '760_R1'] <- 'L60'
# coldata[coldata == '761_R1'] <- 'S61'
# coldata[coldata == '762_R1'] <- 'A62'
# coldata[coldata == '763_R1'] <- 'L63'
# coldata[coldata == '764_R1'] <- 'A64'
# coldata[coldata == '765_R1'] <- 'S65'
# coldata[coldata == '766_R1'] <- 'L66'
# coldata[coldata == '768_R1'] <- 'A68'
# coldata[coldata == '769_R1'] <- 'L69'
# coldata[coldata == 'Ctrl1_R1'] <- 'C1'
# coldata[coldata == 'Ctrl2_R2'] <- 'C2'
# convert column1 with sample names to row.names of coldata
rownames(coldata) <- coldata$Sample_Name
coldata
```

### Adding the groupings by Alina for further Metadata Information

```{r}
# coldata$Epithelial_response <- c("LowInducer", "LowInducer", "HighInducer",
#                                   "HighInducer", "LowInducer", "LowInducer",
#                                   "HighInducer", "HighInducer", "LowInducer",
#                                   "HighInducer", "LowInducer", "HighInducer",
#                                   "LowInducer", "LowInducer", 'NR', 'NR')
# coldata$clinical_outcome <- c('symptomatic', 'symptomatic', 'symptomatic',
#                               'Lethal', 'asymptomatic', 'Lethal', 'symptomatic',
#                               'asymptomatic', 'Lethal', 'symptomatic', 'symptomatic',
#                               'Lethal', 'asymptomatic', 'Lethal', 'NR', 'NR')
# coldata$microcolonies <- c('Low', 'Low', 'Low', 'High', 'Low', 'Low',
#                            'High', 'High', 'Low', 'High', 'Low', 'High', 'Low',
#                            'Low', 'NR', 'NR')
# coldata$ER_microcolonies <- c("LI_LM", "LI_LM", "HI_LM", "HI_HM", "LI_LM", "LI_LM",
#                               "HI_HM", "HI_HM", "LI_LM", "HI_HM", "LI_LM", "HI_HM",
#                               "LI_LM", "LI_LM", 'NR', 'NR')
# coldata$phylogenomic_lineage <- c("EPEC1", "EPEC10", "EPEC9", "EPEC9", "NC", "EPEC5",
#                                   "EPEC8", "NC", "EPEC7", "NC", "EPEC2", "EPEC9",
#                                   "EPEC2", "EPEC2", 'NR', 'NR')
# coldata$phylogroup <- c("B2", "A", "B2", "B2", "B1", "A", "B2", "B2", "B1", "B2", "B1",
#                         "B2", "B1", "B2", 'NR', 'NR')
# coldata$Intimin_Type <- c("alpha", "ND", "lambda", "lambda", "epsilon", "epsilon",
#                           "mu", "lambda", "beta", "kappa", "beta", "alpha", "beta",
#                           "beta", 'NR', 'NR')
# coldata$mixture <- c("Left","Left","Right","Right","Left","Left","Right","Right","Left",
#                       "Right","Left","Right","Left","Left","NR", 'NR')
# coldata$VPosition <- c("Top","Top","Bottom","Top","Bottom","Bottom","Top","Top","Bottom",
#                       "Top","Bottom","Bottom","Bottom","Bottom","NR", 'NR')
```

#### then fix Countsmatrix:

NOTE:

1.  From the manuals the countsData must be a numeric matrix
2.  It is IMPORTANT to keep the names of the genes in the rownames

```{r}
# Readin  countsmatrix
#countsmatrix <-as.matrix(read.csv("~/R/Rtuts/Data/Alina_EPEC_project/counts.csv"))
countsmatrix <- read.csv("~/R/Alina_RNAseq/mixturecounts.csv")
#countsmatrix <- as.data.frame(countsmatrix)
```

```{r}
## Removal of Gender Genes from ENSEMBL ID itself
countsmatrix <- countsmatrix %>% filter(countsmatrix$X != "ENSMUSG00000086503",
                                  countsmatrix$X != "ENSMUSG00000097571",
                                  countsmatrix$X != "ENSMUSG00000086370",
                                  countsmatrix$X != "ENSMUSG00000031329")
nrow(countsmatrix)
#countsmatrix <- as.matrix(countsmatrix)
```

```{r}
#tidying up these names again
#colnames(countsmatrix) <- str_remove_all(colnames(countsmatrix), pattern = "run6_trimmed_|_.bam|_S\\d\\d|_S\\d")
rownames(countsmatrix) <- countsmatrix[,1] #converting first column of gene names into rownames, to be used for sanity check later
# It is IMPORTANT to keep the names of the genes in the rownames
countsmatrix <- subset(countsmatrix, select = - X)#dropping the X column
```
```{r}
# the elements from Sample_Name from coldata must the the colnames of countsmatrix
colnames(countsmatrix) <- coldata$Sample_Name
# Display the column names
colnames(countsmatrix)
```

## Annotating and Exporting ENSEMBL ID into Gene Symbols

Adding genes annotated from ENSEMBL ID to Gene symbols and ENTREZ Id to countsmatrix table. Will be keeping the symbols and entrez columsn to be added later into results table as it is for later use
```{r}
cm_row <- rownames(countsmatrix)
head(cm_row)
# Mapping the ENSEMBL ID to Symbol and ENTREZ ID
symbols <- mapIds(
  org.Mm.eg.db,
  keys = cm_row,
  column = c('SYMBOL'),
  keytype = 'ENSEMBL',
  multiVals = "first"
)
```
```{r}
symbols <- symbols[!is.na(symbols)]
symbols <- symbols[match(rownames(countsmatrix), names(symbols))]
head(symbols, 25)
# Creating a new column called genename and putting in the symbols and entrez columns into count matrix
countsmatrix$genename <- symbols
# Removing all rows with NA values for genenames, so that those rows are filtered out.
countsmatrix <- unique(countsmatrix[rowSums(is.na(countsmatrix)) == 0, ]) # Apply rowSums & is.na
nrow(countsmatrix)
# Moving the ENSEMBL ID from rownames into separate column for itself.
countsmatrix <- tibble::rownames_to_column(countsmatrix, "E_ID")
# Removing the duplicated genes so that then these genes can be made into rownames for countsmatrix
countsmatrix <- distinct(countsmatrix[!duplicated(countsmatrix$genename), ])
```

```{r}
# Now make the ganename column into rownames of count matrix
rownames(countsmatrix) <- countsmatrix[,"genename"]
# Keeping this version of countsmatrix for later use
cm_table <- countsmatrix
# dropping the column E_ID, genenames so that only numeric values are present in it as an input of DESEq Object.
countsmatrix <- subset(countsmatrix, select = -c(genename, E_ID))#
# Changing countsmatrix into Matrix of numeric values so that only numeric values are present in it as an input of DESEq Object.
countsmatrix <- as.matrix(countsmatrix)
class(countsmatrix) <- "numeric"
```


# Calculating CPM Values

```{r}
# as DGEList
dge_er <- DGEList(counts = countsmatrix)
dim(dge_er)
colnames(dge_er)
#dge_er$samples
## calculate norm. factors
nr <- calcNormFactors(dge_er)
## get normalized counts
cpmvalues <- cpm(nr)
cpmvalues_d <- cpm.default(nr)
```

# Differential Gene Expression analysis using DESeq2

Now, construct DESeqDataSet for DGE analysis.

But before that, a sanity check : It is essential to have the name of
the columns in the count matrix in the same order as that in name of the
samples (rownames in coldata).

```{r}
all(rownames(coldata) %in% colnames(countsmatrix))
ncol(countsmatrix) == nrow(coldata)
dim(countsmatrix)
```

## Creating the DESeq Data set Object

```{r}
dds_mixture <- DESeqDataSetFromMatrix(countData = countsmatrix,
                              colData = coldata, 
                              design = ~ condition)
nrow(dds_mixture)
```

```{r}
# Function to save generic plots
saveplot <- function(plot,name ){
  # Function to save the plots
  ggsave(filename = 
           glue('~/R/Alina_RNAseq/mixture/{name}.png'),
       plot = plot,
       dpi = 300,
       width = 10,
       height = 10,
       units = "in")
}
```

## Exploratory Data Analysis and Visualization

### Pre-filtering the dataset

Our count matrix with our DESeqDataSet contains many rows with only
zeros, and additionally many rows with only a few fragments total. In
order to reduce the size of the object, and to increase the speed of our
functions, we can remove the rows that have no or nearly no information
about the amount of gene expression.

Applying the most minimal filtering rule: removing rows of the
DESeqDataSet that have no counts, or only a single count across all
samples. Additional weighting/filtering to improve power is applied at a
later step in the workflow.

```{r}
keep <- rowSums(counts(dds_mixture)) > 1
dds_mixture <- dds_mixture[keep,]
nrow(dds_mixture)
```

### The variance stabilizing transformation

## Applying VST transformation

```{r}
vsd <- vst(dds_mixture, blind = FALSE)
#head(assay(vsd), 3)
colData(vsd)
vsd_coldata <- colData(vsd)
dds_mixture <- estimateSizeFactors(dds_mixture)
dds_mixture
```

## Sample Distances

useful first step in an RNA-seq analysis is often to assess overall
similarity between samples:

1.  Which samples are similar to each other, which are different?
2.  Does this fit to the expectation from the experiment's design?

### Euclidean Distance between samples

dist to calculate the Euclidean distance between samples - useful for
ONLY normalized data. To ensure we have a roughly equal contribution
from all genes, we use it on the VST data.

```{r}
sampleDists <- dist(t(assay(vsd)))
sampleDistMatrix <- as.matrix( sampleDists )
rownames(sampleDistMatrix) <- vsd$Sample_Name
colnames(sampleDistMatrix) <- vsd$Sample_Name
colors <- colorRampPalette( rev(brewer.pal(9, "RdYlBu")) )(255)
(EuclideanDistanceHeatmap <- pheatmap(sampleDistMatrix,
         clustering_distance_rows = sampleDists,
         clustering_distance_cols = sampleDists,
         main = "Sample-to-Sample Euclidean Distance of Mixture of Samples",
         #col = colors
         ))
```

```{r}
#saveplot(EuclideanDistanceHeatmap, "EuclideanDistanceHeatmap")
```

### Poisson Distance between Samples

```{r}
poisd <- PoissonDistance(t(counts(dds_mixture))) # raw counts or unnormalised data
samplePoisDistMatrix <- as.matrix( poisd$dd )
rownames(samplePoisDistMatrix) <- dds_mixture$Sample_Name
colnames(samplePoisDistMatrix) <- dds_mixture$Sample_Name
colors <- colorRampPalette( rev(brewer.pal(9, "RdYlBu")) )(255)
(poisson_dist_plot <- pheatmap(samplePoisDistMatrix,
         clustering_distance_rows = poisd$dd,
         clustering_distance_cols = poisd$dd,
         main = "Sample-to-Sample Poisson Distance of Mixture of Samples",
         col = colors))
         
```

```{r}
#saveplot(poisson_dist_plot, "poisson_dist_plot")
```

# PCA Plot
```{r}
### Functions for Plot aethetics and saving PCA Plots
# color_values <- c("red", "red","red","red", "black","black","red","red", "red",
#                   "red", "red","red","red", "red", "red" ,"blue")
color_values <- c("red", "red","blue","blue","black","black","green","green", "gray",
                  "gray", "brown","brown")
# the basic set of common aesthetic settings for PCA plots, 
theme.my.own <- list(theme_bw() ,
                      geom_point(size = 3),
                      coord_fixed() ,
                      scale_y_continuous(breaks = seq(-20, 20, 5), 
                                         sec.axis = sec_axis(~. *1, 
                                                             labels = NULL,
                                                             breaks = NULL)) ,
                      scale_x_continuous(breaks = seq(-20, 20, 5), 
                                         sec.axis = sec_axis(~. *1, 
                                                             labels = NULL,
                                                             breaks = NULL)) ,
                      theme_classic() ,
                      geom_hline(yintercept = 0, color = "gray", size = 1) ,
                      geom_vline(xintercept = 0, color = "gray", size = 1) ,
                      theme(text = element_text(size = 15),
                            axis.text = element_text(size = 15),
                            legend.position = "right",
                            aspect.ratio = 1) ,
                      #geom_text(size = 4, hjust = 0, vjust = 0)
                      geom_text_repel(size = 5,min.segment.length = 0.5)
  )
```

## Calculating all PCA Values

```{r}
plotPCA_local = function(object,
                         intgroup = "condition",
                         ntop = 500,
                         returnData = TRUE,
                         nPC = 4)
{
  # calculate the variance for each gene
  rv <- rowVars(assay(object))
  ntop <- 500
  
  # select the ntop genes by variance
  select <- order(rv, decreasing = TRUE)[seq_len(min(ntop, length(rv)))]
  
  # perform a PCA on the data in assay(x) for the selected genes
  pca <- prcomp(t(assay(object)[select, ]))
  #summary(pca)
  
  # the contribution to the total variance for each component
  percentVar <- pca$sdev ^ 2 / sum(pca$sdev ^ 2)
  
  if (!all(intgroup %in% names(colData(object)))) {
    stop("the argument 'intgroup' should specify columns of colData(dds)")
  }
  
  intgroup.df <-
    as.data.frame(colData(object)[, intgroup, drop = FALSE])
  
  # add the intgroup factors together to create a new grouping factor
  group <- if (length(intgroup) > 1) {
    factor(apply(intgroup.df, 1, paste, collapse = ":"))
  } else {
    colData(object)[[intgroup]]
  }
  
  # assembly the data for the plot
  d <- cbind(pca$x[, seq_len(min(nPC, ncol(pca$x))), drop = FALSE],
             data.frame(group = group, intgroup.df, name = colnames(object)))
  
  if (returnData) {
    attr(d, "percentVar") <- percentVar[1:nPC]
    # l <- list(pca,d)
    # return(l)
     return(d)
  }
}
  
```

## PCA Plot with VST Data

### Function for calculating percentvar

```{r}
percentvar_calculation <- function(pcaData_variable){
  # function to calculate percentvar for different variables
  percentvar_variable <- round(100 * attr(pcaData_variable, "percentVar"), digits = 3 )
  return(percentvar_variable)
}
savingFunction <- function(plotname, metadatacolumn){
  # Function to save the PCA plots
  ggsave(filename = 
           glue('~/R/Alina_RNAseq/mixture/PCAplot_mixture_{metadatacolumn}.png'),
       plot = plotname,
       dpi = 300,
       width = 10,
       height = 10,
       units = "in")
}
```

```{r}
pcaData_mixture <- plotPCA_local(vsd, intgroup = c("condition","Sample_Name"),returnData = T)
pcaData_mixture
percentVar_mixture <- percentvar_calculation(pcaData_mixture)
```

```{r }
percentVar_mixture
```
```{r fig.width=10, fig.height=10}
(PCAplot_vst <- ggplot(pcaData_mixture,
                      aes(x = PC1,
                          y = PC2,
                          color = Sample_Name,
                          label = Sample_Name)) +
        xlab(paste0("PC1: ", percentVar_mixture[1], "% variance")) +
        ylab(paste0("PC2: ", percentVar_mixture[2], "% variance")) +
        ggtitle("PCA Plot - Mixture") +
        scale_colour_manual(values = color_values) +
        theme.my.own )
savingFunction(PCAplot_vst, "condition")
```
## PCA Plot for PC2 vs PC3

```{r}
(PCAplot_vst23 <- ggplot(pcaData_mixture,
                      aes(x = PC2,
                          y = PC3,
                          color = Sample_Name,
                          label = Sample_Name)) +
        xlab(paste0("PC2: ", percentVar_mixture[2], "% variance")) +
        ylab(paste0("PC3: ", percentVar_mixture[3], "% variance")) +
        ggtitle("PCA Plot - Mixture") +
        scale_colour_manual(values = color_values) +
        theme.my.own )
# ggsave(filename = '~/R/Alina_RNAseq/mixture/PCAplot23_mixturevsControl.png',
#        plot = PCAplot_vst23,
#        dpi = 300,
#        width = 10,
#        height = 10,
#        units = "in")
```


## PCA Plot for PC3 vs PC4

```{r}
(PCAplot_vst34 <- ggplot(pcaData_mixture,
                      aes(x = PC3,
                          y = PC4,
                          color = Sample_Name,
                          label = Sample_Name)) +
        xlab(paste0("PC3: ", percentVar_mixture[3], "% variance")) +
        ylab(paste0("PC4: ", percentVar_mixture[4], "% variance")) +
        ggtitle("PCA Plot - Mixture") +
        scale_colour_manual(values = color_values) +
        theme.my.own )
# ggsave(filename = '~/R/Alina_RNAseq/mixture/PCAplot34_mixturevsControl.png',
#        plot = PCAplot_vst34,
#        dpi = 300,
#        width = 10,
#        height = 10,
#        units = "in")
```


```{r}
# calculate the variance for top 500 gene
rv <- rowVars(assay(vsd))
ntop <- 500
# select the ntop genes by variance
select <- order(rv, decreasing = TRUE)[seq_len(min(ntop, length(rv)))]
df1 <- t(assay(vsd)[select,])
```
```{r}
res.pca <- PCA(df1,  graph = FALSE, scale.unit = FALSE)
summary.PCA(res.pca)
```
```{r}
# Visualize eigenvalues/variances
fviz_screeplot(res.pca, addlabels = TRUE)
```
```{r}
library("factoextra")
eig.val <- get_eigenvalue(res.pca)
eig.val
```

## Genes + PCA Biplots

```{r}
fviz_pca_biplot(res.pca, repel = TRUE,
                gradient.cols = c("pink", "blue", "yellow", "green", "red", "black"))
```

```{r fig.height=10, fig.width=10}
heat.colors <- brewer.pal(6, "RdYlBu")
fviz_pca_var(res.pca, col.var = "contrib", repel = TRUE,
             gradient.cols = c("Gray", "blue", "yellow","orange", "green", "red", "black"),
             )
```

```{r}
# Contributions of variables to PC2
fviz_pca_contrib(res.pca, choice = "var", axes = 2, top = 25)
```
```{r}
# Contributions of variables to PC1
fviz_contrib(res.pca, choice = "var", axes = 1, top = 25)
```


```{r}
sessionInfo()
```