MaxQuant_Summary_ExpressionAtlas_LabelFree_Rmarkdown.Rmd

---
title: "Summary of reanalysed MaxQuant output exported into Expression Atlas"
output: pdf_document
---




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

library(plyr)
library(dplyr)
library(tidyr)
library(knitr)
library(stringr)
library(mygene)
library(ggplot2)
library(gtools)
library(reshape2)
library(Biobase)
library(finalfit)
library(data.table)
library(tibble)
library(kableExtra)


dir <- "/Users/ananth/Documents/MaxQuant_Bechmarking/Human/Synapse-AD/ACT_DorsoLateralPreFrontalCortex/"

tmp  <- read.table( paste(dir,"proteinGroups.txt", sep="") , quote = "\"", header = TRUE, sep = "\t", stringsAsFactors = FALSE, comment.char = "#")
idf  <- read.table( paste(dir,"E-PROT-51.idf.txt", sep="") , quote = "\"", header = TRUE, sep = "\t", stringsAsFactors = FALSE, comment.char = "#")

PI_name <-tibble::enframe(idf[idf$MAGE.TAB.Version == "Person Last Name" | idf$MAGE.TAB.Version == "Person First Name",3])
PI_email <- tibble::enframe(idf[idf$MAGE.TAB.Version == "Person Email",3])
pubmed_id <- tibble::enframe(idf[idf$MAGE.TAB.Version == "PubMed ID",2])
pride_id <- tibble::enframe(idf[idf$MAGE.TAB.Version == "Comment[SecondaryAccession]",2])
pride_link <- tibble::enframe(paste("https://www.ebi.ac.uk/pride/archive/projects/", pride_id$value, sep=""))


###### Functions ############
#####
# Normalise iBAQ values
# FOT normalisation
# define FOT normalisation function
# FOT stands for Fraction Of Total. In this normalisation method each protein iBAQ intensity value is scaled to the total amount of signal in a given MS run (column) and transformed to parts per billion (ppb)

fot.normalise <- function(x){
  data.sum <-   apply(x, 2, function(y){sum(y, na.rm=TRUE)})
  ##### do ppm normalisation
  x.mat <- as.matrix(x)
  x.mat.ppb <- apply(x.mat, 2, function(i) i/sum(i, na.rm = T) * 1000000000 )
  x.mat.ppb <- as.data.frame(x.mat.ppb)
  colnames(x.mat.ppb) <- paste("ppb.", colnames(x.mat.ppb), sep = "")
  return(x.mat.ppb)
}


#####
# Counting number of proteins in each sample
# Since there is no direct protein count for each sample in MaxQuant output
# we look at intensity coloumn for each sample and if the intensity value is 0, then the protein count is 0
# for non zero intensity count the protein count for that sample is the same as in teh coloumn "Number.of.proteins".
# https://mgimond.github.io/ES218/Week03a.html

Proteincount_per_sample <- function(df){
  Intensity_data_cols <- c(grep("Intensity.", colnames(df) ) )
  Intensity <- df[,c("Protein.IDs","Number.of.proteins")]
  
  Proteincount_persample <- Intensity
  
  Proteincount_persample$Sprot_counts <- NA
  
  for(i in 1:nrow(Proteincount_persample)){  
    y <- strsplit(Proteincount_persample[ i, "Protein.IDs"], split = ";")
    y <- unique(gsub("-\\d+", "", y[[1]], perl=TRUE)) # remove isoform number and consider unique entries
    Proteincount_persample[i, "Sprot_counts"] <- sum(str_count(y, "sp\\|")) # only count SwissProt entries
  }
  
  Proteincount_persample <- cbind(Proteincount_persample, df[,Intensity_data_cols[1]:tail(Intensity_data_cols, n=1)])
  
  Proteincount_persample <- mutate_at(Proteincount_persample, vars(colnames(Proteincount_persample[,c(-1,-2, -3)])), 
                                      list(~ ifelse( . == 0, 0, Proteincount_persample$Sprot_counts)))
  
  setnames(Proteincount_persample, old = c(-1,-2,-3), new = gsub("Intensity", "Protein.count",colnames(Proteincount_persample[,c(-1,-2,-3)]),perl=TRUE))
  
  Proteincount_sum_persample <- data.frame(lapply(Proteincount_persample[,c(-1,-2,-3)], function(y){sum(y)}))
  return(Proteincount_sum_persample)
}



#####
# Counting number of peptides in each sample

Peptide_count_per_sample <- function(df){
  Peptide_data_cols <- c(grep("", colnames(df)))
  Peptidecount_sum_persample <- data.frame(lapply(df[,Peptide_data_cols[1]:tail(Peptide_data_cols, n=1)], function(y){sum(y)}))
  Peptidecount_cnames <- colnames(Peptidecount_sum_persample)
  Peptidecount_long <- gather(Peptidecount_sum_persample, Samples, Peptide_count_sum, Peptidecount_cnames[1]:tail(Peptidecount_cnames, n=1), factor_key=TRUE)
  return(Peptidecount_long)
}



###### Pre-processing #######

# count number of samples and protein groups (before filtering)
number_of_proteingroups <- nrow(tmp)
number_of_contaminants_pre <- nrow(tmp[ tmp[, "Potential.contaminant"] == "+", ])
number_of_reversedecoys_pre <- nrow(tmp[ tmp[, "Potential.contaminant"] == "+", ])
number_of_samples_pre <- ncol(tmp[ ,c(grep("Peptides.", colnames(tmp))) ])
number_of_peptides_pre <- sum(tmp$Peptides)
number_of_unique_peptides_pre <- sum(tmp$Unique.peptides)
number_of_unique_gene_mappings_pre <- NA

#Count number of SwissProt entries without isoforms (before filtering)
tmp$Sprot_counts <- "NA"

for(i in 1:nrow(tmp)){
    
  y <- strsplit(tmp[ i, "Protein.IDs"], split = ";")
  y <- unique(gsub("-\\d+", "", y[[1]], perl=TRUE)) # remove isoform number and consider unique entries
  
  tmp[i, "Sprot_counts"] <- sum(str_count(y, "sp\\|"))
}

#Count number of SwissProt entries without isoforms (before filtering)
Swissprot_sum_pre <- sum(as.numeric(as.character(tmp$Sprot_counts)))

table1 <- rbind(number_of_samples_pre, number_of_contaminants_pre, number_of_reversedecoys_pre, Swissprot_sum_pre, number_of_peptides_pre, number_of_unique_gene_mappings_pre, number_of_unique_peptides_pre)
 
###### Post-processing ######
# Apply filter 1
tmp <- tmp[ tmp[, "Reverse"] != "+", ]
tmp <- tmp[ tmp[, "Potential.contaminant"] != "+", ]
tmp <- tmp[ tmp[, "Peptides"] > 1, ]
number_of_samples_post <- ncol(tmp[ ,c(grep("Peptides.", colnames(tmp))) ])

EXP_TYPE <- "MS1-quant"

if(EXP_TYPE == "MS1-quant"){
  tmp_iBAQs_pre_normalised <- tmp[ ,c(2, grep("iBAQ.", colnames(tmp))) ]
  
  Majority.protein.IDs <- tmp$Majority.protein.IDs
  tmp_nonids <- tmp_iBAQs_pre_normalised[,-1]
  tmp_iBAQs_ppb_normalised <- fot.normalise(tmp_nonids)
  tmp_iBAQs_ppb_normalised <- data.frame( cbind(Majority.protein.IDs, tmp_iBAQs_ppb_normalised, stringsAsFactors = FALSE) )
}

tmp_iBAQs_ppb_normalised[tmp_iBAQs_ppb_normalised == 0] <- NA
tmp_iBAQs_ppb_normalised[tmp_iBAQs_ppb_normalised == "NaN"] <- NA

tmp_iBAQs_ppb_normalised$"ENSG" <- "NA"
tmp_iBAQs_ppb_normalised$"Gene.Name" <- "NA"
tmp_iBAQs_ppb_normalised$"Gene.Symbol" <- "NA"
tmp_iBAQs_ppb_normalised$"unique.gene.count" <- "NA"
tmp_iBAQs_ppb_normalised$Peptides <- tmp$Peptides
tmp_iBAQs_ppb_normalised$Unique.peptides <- tmp$Unique.peptides
tmp_iBAQs_ppb_normalised$Potential.contaminant <- tmp$Potential.contaminant
tmp_iBAQs_ppb_normalised$Reverse <- tmp$Reverse

# Map UniProt Accessionto ENSEMBL Gene ID/Symbol
for(i in 1:nrow(tmp_iBAQs_ppb_normalised)){
    
    x <- data.frame(strsplit(tmp_iBAQs_ppb_normalised[ i, "Majority.protein.IDs"], split = ";"), stringsAsFactors = FALSE)
    x_temp <- regmatches(x[,1],regexpr("[OPQ][0-9][A-Z0-9]{3}[0-9]|[A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2}", x[,1]))
    #x[,1] <- gsub("sp\\||tr\\|", "", x[,1], perl=TRUE)
    #x[,1] <- gsub("\\|.*", "", x[,1], perl=TRUE)
    #x[,1] <- gsub("-\\d+", "", x[,1], perl=TRUE) # remove isoform number
    
    f = file()
    sink(file=f)
    a <- tryCatch(queryMany(x_temp, scopes="uniprot", fields=c("ensembl.gene", "symbol", "name"), species="human"), error = function(e) {print(0)})
    sink()
    close(f)
                 # DFrame not DataFrame 
    if (class(a)=="DFrame" | class(a) == "DataFrame"){
    tmp_iBAQs_ppb_normalised[ i, "ENSG"] <- paste( unique(unlist(a$ensembl.gene[!is.na(a$ensembl.gene)])), collapse = ";")
    tmp_iBAQs_ppb_normalised[ i, "Gene.Name"] <- paste( unique(unlist(a$name[!is.na(a$name)])), collapse = ";")
    tmp_iBAQs_ppb_normalised[ i, "Gene.Symbol"] <- paste( unique(unlist(a$symbol[!is.na(a$symbol)])), collapse = ";")
    temp_symb <- tmp_iBAQs_ppb_normalised[i,"Gene.Symbol"]
    tmp_iBAQs_ppb_normalised[ i , "unique.gene.count"] <- str_count(unique(temp_symb), ";")+1
  }
}

# Apply Filter 2
# remove protein groups that have no mapping to an ENSG gene IDs
tmp_iBAQs_ppb_normalised <- tmp_iBAQs_ppb_normalised[tmp_iBAQs_ppb_normalised$ENSG != "" ,]
tmp_iBAQs_ppb_normalised <- tmp_iBAQs_ppb_normalised[tmp_iBAQs_ppb_normalised$ENSG != "NA" ,]
tmp_iBAQs_ppb_normalised <- tmp_iBAQs_ppb_normalised[tmp_iBAQs_ppb_normalised$unique.gene.count == 1, ]
tmp_iBAQs_ppb_normalised <- tmp_iBAQs_ppb_normalised[tmp_iBAQs_ppb_normalised[, "Peptides"] > 1, ]

number_of_peptides_post <- sum(tmp_iBAQs_ppb_normalised$Peptides)
number_of_unique_peptides_post <- sum(tmp_iBAQs_ppb_normalised$Unique.peptides)
number_of_contaminants_post <- nrow(tmp_iBAQs_ppb_normalised[ tmp_iBAQs_ppb_normalised[, "Potential.contaminant"] == "+", ])
number_of_reversedecoys_post <- nrow(tmp_iBAQs_ppb_normalised[ tmp_iBAQs_ppb_normalised[, "Reverse"] == "+", ])
number_of_unique_gene_mappings_post <- length(unique(tmp_iBAQs_ppb_normalised$Gene.Symbol))

# remove all protein groups that map to multiple ENSG gene IDs (this removes a lot of proteins) - the reasoning to remove these cases is that we cannot establish for sure which gene is contributing the signal to the protein abundance; all genes contribute equally or one gene is a majority? 
tmp_iBAQs_ppb_normalised <- tmp_iBAQs_ppb_normalised[ grep(";", tmp_iBAQs_ppb_normalised$ENSG, invert = TRUE) , ]

tmp_iBAQs_ppb_normalised <- aggregate(tmp_iBAQs_ppb_normalised[ , 2:(ncol(tmp_iBAQs_ppb_normalised)-8)], list("Gene ID" = tmp_iBAQs_ppb_normalised$ENSG, "Gene.Symbol" = tmp_iBAQs_ppb_normalised$Gene.Symbol), sum, na.rm =TRUE)




# iBAQs

tmp_iBAQs_ppb_normalised <- tmp_iBAQs_ppb_normalised[ ,c(1, grep("iBAQ.", colnames(tmp_iBAQs_ppb_normalised))) ]

cnames_iBAQs_ppb <- colnames(tmp_iBAQs_ppb_normalised)


tmp_iBAQs_ppb_table <- gather(tmp_iBAQs_ppb_normalised, Samples, iBAQ, cnames_iBAQs_ppb[2]:tail(cnames_iBAQs_ppb, n=1), factor_key=TRUE)
tmp_iBAQs_ppb_table$Samples <- gsub("iBAQ.|ppb.iBAQ.", "", tmp_iBAQs_ppb_table$Samples, perl=TRUE)
tmp_iBAQs_ppb_table <- tmp_iBAQs_ppb_table[str_order(tmp_iBAQs_ppb_table$Samples, numeric=TRUE),]
save(tmp_iBAQs_ppb_table, file = paste(dir,"iBAQ_plot_ExAtlas.rda", sep=""))


#####
# Counting number of proteins in each sample (post-processed)

Proteincount_persample_post <- Proteincount_per_sample(tmp)
Proteincount_cnames_post <- colnames(Proteincount_persample_post)
proteincount_table_post <- gather(Proteincount_persample_post, Samples, Protein_count_sum, Proteincount_cnames_post[1]:tail(Proteincount_cnames_post, n=1), factor_key=TRUE)
proteincount_table_post <- proteincount_table_post[str_order(proteincount_table_post$Samples, numeric=TRUE),]
proteincount_table_post$Samples <- gsub("Protein.count.", "", proteincount_table_post$Samples, perl=TRUE) 
save(proteincount_table_post, file = paste(dir,"proteincount_plot_ExAtlas.rda", sep=""))


#####
#Count number of SwissProt entries without isoforms (after filtering)
Swissprot_sum_post <- sum(as.numeric(as.character(tmp$Sprot_counts)))


#####
# Counting number of peptides in each sample
peptidecount_persample <- tmp[, c(grep("Peptides.", colnames(tmp)))]
peptidecount_table <- Peptide_count_per_sample(peptidecount_persample )
peptidecount_table <- peptidecount_table[str_order(peptidecount_table$Samples, numeric=TRUE),]
peptidecount_table$Samples <- gsub("Peptides.", "", peptidecount_table$Samples, perl=TRUE) 
save(peptidecount_table, file = paste(dir,"peptidecount_plot_ExAtlas.rda", sep=""))



##############
## For Plots

# Protein identification overlap
# counting number of proteins that are found across different samples.
Protein_overlap <- function(df){
  Intensity_data_cols <- c(grep("Intensity.", colnames(df) ) )
  Intensity <- df[,c("Protein.IDs","Number.of.proteins")]
  
  Intensity <- cbind(Intensity, df[,Intensity_data_cols[1]:tail(Intensity_data_cols, n=1)])
  
  Protein_overlap <- Intensity
  Protein_overlap <- mutate_at(Protein_overlap, vars(colnames(Protein_overlap[,c(-1,-2)])), 
                                      list(~ ifelse( . == 0, 0, 1)))
  
  Protein_overlap_all <- data.frame(Protein.overlap = rowSums(Protein_overlap[,c(-1,-2)]))
  return(Protein_overlap_all)
}

Protein_overlap_all <- Protein_overlap(tmp)
tmp <- cbind(tmp, Protein_overlap_all)

#dplyr::count gives error hence using plyr::count
protein_overlap_freq <- tmp %>% plyr::count('Protein.overlap')
save(protein_overlap_freq, file = paste(dir,"protein_overlap_plot_ExAtlas.rda", sep=""))


table2 <- rbind(number_of_samples_post, number_of_contaminants_post, number_of_reversedecoys_post, Swissprot_sum_post, number_of_peptides_post, number_of_unique_gene_mappings_post, number_of_unique_peptides_post)


#### Summary table
table <- cbind(table1, table2)
rownames(table) <- c("Number of samples", "Number of potential contaminants•", "Number of reverse decoys^", "Number of identified proteins†", "Total number of mapped peptidesª", "Protein groups mapped to unique gene id¢" , ".... Number of mapped unique peptides§")
colnames(table) <- c("Pre-processed", "Post-processed*")
save(table, file = paste(dir,"table_ExAtlas.rda", sep=""))



#Correlation between samples based on normalised iBAQ
cor_values <- cor(tmp_iBAQs_ppb_normalised[-1])
colnames(cor_values) <- gsub("ppb.iBAQ.", "", colnames(cor_values), perl=TRUE)
rownames(cor_values) <- gsub("ppb.iBAQ.", "", rownames(cor_values), perl=TRUE)
row.order <- hclust(dist(cor_values))$order
col.order <- hclust(dist(t(cor_values)))$order
cor_values_clustered <- cor_values[row.order, col.order]
melted_cormat <- melt(cor_values_clustered)
save(melted_cormat, file = paste(dir,"correlationmat_ExAtlas.rda", sep=""))

```
  
<!--
Section to print in PDF document
-->

PRIDE dataset identifier:\hspace{10pt}   `r pride_id$value`  
PRIDE dataset URL:\hspace{29pt}          `r pride_link$value`  
Lab Head:\hspace{81pt}                   `r pride_id$PI_name`  
E-mail:\hspace{94pt}                     `r pride_id$PI_email`  
Affiliation:\hspace{80pt}                Goizueta Alzheimer's Disease Research Center, Emory University School of Medicine, Atlanta, GA, USA  
Original dataset submitter:\hspace{7pt}  Johnson, Erik C. B.   
E-mail:\hspace{94pt}                     erik.c.b.johnson@emory.edu  
PubMed ID:\hspace{73pt}                 `r pubmed_id$value`  

Quantification method:\hspace{25pt}      Label-free (baseline)  
Search database:\hspace{53pt}            Human Reference Proteome (UniProt, May 2019. 95,915 sequences)  
Contaminant database:\hspace{25pt}       MaxQuant contaminants database (conf/contaminants.fasta)  
Analysis software:\hspace{48pt}          MaxQuant v1.6.3.4  
Operating system:\hspace{47pt}           Red Hat Enterprise Linux Server


 

# Summary table   

MaxQuant output before and after processing.  
The submitted original '.raw' files are run through MaxQuant; the output (pre-processed) intensities are then normalised, proteins mapped to Ensembl gene IDs and filtered results (post-processed) are uploaded to Expression Atlas.  
```{r table, echo = FALSE}
load(paste(dir,"table_ExAtlas.rda", sep=""))
kable(table, row.names=TRUE, booktabs = T) %>%
kable_styling(position="left")  
```

\setlength{\leftskip}{0.5cm}* Data show in Expression Atlas.  
\setlength{\leftskip}{0.5cm}• The total number of protein groups found to be a commonly occurring contaminant.  
\setlength{\leftskip}{0.5cm}**^** The total number of protein groups with a peptide derived from the reversed part of the decoy database.  
\setlength{\leftskip}{0.5cm}**†** The total number of non-isoform SwissProt proteins within th protein group, to which at least 2 or more peptides from each sample are mapped to.  
\setlength{\leftskip}{0.5cm}**ª** Sum of peptides that are mapped across all protein groups.   
\setlength{\leftskip}{0.5cm}**¢** The total number of protein groups which are mapped to an unique Ensembl Gene ID.  
\setlength{\leftskip}{0.5cm}§ The total number of unique peptides associated with the protein group (i.e. these peptides are not shared with another protein group).  

\setlength{\leftskip}{0pt}

# Post-processing filters applied:
(i) Remove reverse decoys.
(ii) Remove potential contaminants.
(iii) Include protein groups to which 2 or more unique peptides are mapped.
(iv) Include protein groups wherein all protein IDs within are mapped to an unique Ensembl Gene ID.

    
# Normalisation method:
Fraction Of Total (FOT): Each protein iBAQ intensity value is scaled to the total amount of signal in a given MS run (column) and transformed to parts per billion (ppb)  


```{r tmp_iBAQs_ppb_table, fig.height = 4, echo = FALSE, message=FALSE}
# iBAQ plot after normalisation
load(paste(dir,"iBAQ_plot_ExAtlas.rda", sep=""))
suppressWarnings(print(
ggplot(tmp_iBAQs_ppb_table, aes(x=factor(tmp_iBAQs_ppb_table$Samples, levels = unique(tmp_iBAQs_ppb_table$Samples)),
                                y = iBAQ, colour = Samples))+
  geom_boxplot(na.rm=TRUE)+
  theme_bw()+
  theme(legend.position="none")+
  xlab("Samples")+
  scale_y_log10()+
  scale_x_discrete(labels = function(x) str_wrap(x, width = 10))+
  theme(axis.text.x = element_text(angle = 90, hjust = 1))+
  #coord_flip()+
  ggtitle("FOT normalised iBAQ values")
))
```
Figure 1. Boxplots with distribution of iBAQ values for each sample after FOT normalisation.  


```{r protein_overlap_plot, fig.height = 3, fig.width= 6, echo = FALSE}
load(paste(dir,"protein_overlap_plot_ExAtlas.rda", sep=""))
ggplot(protein_overlap_freq[protein_overlap_freq$Protein.overlap != 0, ], aes(x=factor(Protein.overlap), y = freq))+
  geom_col()+
  geom_text(stat = 'identity',aes(label =freq, vjust = -0.2), size=2)+
  theme_bw()+
  xlab("Number of samples") +
  ylab("Number of protein groups")+
  theme(legend.position="none")+
  ggtitle("Protein groups commonly identified across samples")
```
Figure 2. Protein overlap.  Indicates the number of protein groups that were identified across different samples. Protein groups were counted as present in a sample when the sample had registered intensity.  
  
  
```{r Proteincount_plot, fig.height = 3, echo = FALSE}
# Protein count per sample after filtering
load(paste(dir,"proteincount_plot_ExAtlas.rda", sep=""))
ggplot(proteincount_table_post, aes(x=factor(Samples, levels = Samples), 
                                    y = Protein_count_sum, fill=Samples))+
  geom_col(position="dodge", alpha=0.5)+
  geom_text(stat = 'identity',aes(label =Protein_count_sum, vjust = -0.2), size=2)+
  theme_bw()+
  theme(legend.position = "none")+
  xlab("Samples")+
  ylab("Total number of proteins across\nall protein groups")+
  scale_x_discrete(labels = function(x) str_wrap(x, width = 10))+
  theme(axis.text.x = element_text(angle = 90, hjust = 1))+
  theme(legend.position="none")+
  ggtitle("Total number of proteins identified in each sample")

```
Figure 3.  Protein counts in each sample. The total number of proteins (SwissProt non-isoforms) from all protein groups to which at least 2 or more unique peptides from each sample are mapped to.   
  
\vspace{20pt} 
```{r Peptidecount_plot, fig.height = 3, echo = FALSE}
# Peptide count per sample
load(paste(dir,"peptidecount_plot_ExAtlas.rda", sep=""))
ggplot(peptidecount_table, aes(x=factor(Samples, levels = Samples), 
                               y = Peptide_count_sum, fill=Samples))+
  geom_col(position="dodge", alpha=0.5)+
  geom_text(stat = 'identity',aes(label =Peptide_count_sum, vjust = -0.2), size=2)+
  theme_bw()+
  theme(legend.position = "none")+
  xlab("Samples")+
  ylab("Total number of peptides across\nall protein groups")+
  scale_x_discrete(labels = function(x) str_wrap(x, width = 10))+
  theme(axis.text.x = element_text(angle = 90, hjust = 1))+
  theme(legend.position="none")+
  ggtitle("Number of peptides mapped per sample")

```
Figure 4.  Peptide counts in each sample. The total number of peptides that are mapped across all protein groups from each sample.  

   
```{r Correlation_plot, fig.height = 8, fig.width = 8, echo = FALSE}
# Correlation between samples
load(paste(dir,"correlationmat_ExAtlas.rda", sep=""))
ggplot(melted_cormat, aes(x=Var1, y=Var2, fill=value))+
  geom_tile()+
  theme_bw()+
  xlab("")+
  ylab("")+
  theme(axis.text.x = element_text(angle = 90, hjust = 1))+
  #guides(fill=guide_legend(title=expression(paste("Pearson R"^2))))+
  guides(fill=guide_legend(title="Pearson r"))+
  ggtitle("Clustered map of correlation between samples")
  
```
  
  
Figure 5.  Correlation between samples. The pairwise Pearson correlation was calculated between normalised intensities (iBAQs) of each sample and clustered heirarchically.  
  

 <!--```{r Correlation_plot1, fig.height = 5, fig.width = 6, echo = FALSE}
# Correlation between samples clustered
load(paste(dir,"correlationmat1_ExAtlas.rda", sep=""))
heatmap(cor_values , Colv=T, scale='none', cexRow=1, cexCol=1)
``` -->

# Glossary
The descriptions of the terms used in here are taken from MaxQuant documentation.  

**Reverse decoy**: This particular protein group contains no protein, made up of at least 50% of the peptides of the leading protein, with a peptide derived from the reversed part of the decoy database. These are removed for further data analysis. The 50% rule is in place to prevent spurious protein hits to erroneously flag the protein group as reverse.  

**Potential contaminant**: This particular protein group was found to be a commonly occurring contaminant. These are removed for further data analysis.    

**Peptides**: The total number of peptide sequences associated with the protein group (i.e. for all the proteins in the group).  

**Unique peptides**: The total number of unique peptides associated with the protein group (i.e. these peptides are not shared with another protein group).