# Load libs functions
library("fgsea")  
library("DESeq2")
library("ggplot2")
library("pheatmap")
library("biomaRt")
source("AnalyseMetadataCorrelationBulk.R")

# Counts and metadata
  rawCounts <- data.frame(read.table("Pln/Aso/readcounts_raw.txt", header=T, row.names=1, stringsAsFactors=F))
  metadata <- data.frame(read.table("Pln/Aso/sample_annot.txt", sep='\t', header=T, row.names=1))

# Differential gene expression  
  dds <- DESeqDataSetFromMatrix(countData=rawCounts[,rownames(metadata)], colData=metadata, design= ~ Condition)
  dds$Condition <- relevel(dds$Condition, ref='8w_wt_unt')
  dds <- DESeq(dds)
  #DE Lists
  res.hom.over.wt <- as.data.frame(results(dds, contrast=c("Condition", '8w_hom_sham', '8w_wt_unt')))
  res.treatment.over.sham <- as.data.frame(results(dds, contrast=c("Condition", '8w_hom_tr', '8w_hom_sham')))
  res.treatment.over.wt <-as.data.frame(results(dds, contrast=c("Condition", '8w_hom_tr', '8w_wt_unt')))
 
# Principle component analysis
  rld <- rlog(dds, blind=FALSE)
  pcadata <- plotPCA(rld, intgroup=colnames(metadata), returnData=T)
  pca <-  ggplot(pcadata, aes(x=PC1, y=PC2, group = Condition)) + 
  geom_point(aes(color=Condition))

# Gene set enrichment analysis
  genesForGSEA <- res.hom.over.wt
  outDir <- 'hom.over.wt'
  # Convert to human orthologous genes
  human = useMart("ensembl", dataset = "hsapiens_gene_ensembl")
  mouse = useMart("ensembl", dataset = "mmusculus_gene_ensembl")
  ortho_table <- getLDS(attributes = c('ensembl_gene_id'),filters = 'ensembl_gene_id', values = row.names(genesForGSEA), mart = mouse,attributesL = c("hgnc_symbol"), martL = human, uniqueRows = TRUE)
  genesForGSEA <- merge(genesForGSEA, ortho_table, by.x="row.names", by.y=colnames(ortho_table)[1], sort=F, all.x=F, all.y=F)
  # run GSEA
  gmtFiles <- list.files('Database/MSigdb/v6.2/symbol/')
  fgseaRes <- list()
  for(j in 1:length(gmtFiles)) {
      rankedgenesForGSEA <- genesForGSEA[order(genesForGSEA[,'stat'], decreasing = T ),]
      rankedGenes <- rankedgenesForGSEA[,'stat']
      names(rankedGenes) <- rankedgenesForGSEA$HGNC.symbol
      pathwaysForGSEA <- gmtPathways(file(paste0('/Users/k.boogerd/Documents/DATA/R/Scripts/GSEA/test/',gmtFiles[j])))
      fgseaRes[[gmtFiles[j]]] <- fgsea(pathways = pathwaysForGSEA, stats = rankedGenes, minSize=15,maxSize=500,nperm=10000,gseaParam = 1)
      }

# Correlation analysis
  corrLVEF <- analyseMetaDataCorrelationBulk(
    scsMetadata=metadata, 
    dfCounts=counts(dds, normalized=TRUE), 
    parameterName = 'LVEF',
    percentage=50, 
    correlationMethod = "spearman")