See Issue #14, Issue #18, Issue #19

Author: gabrielodom

R/CloseBySingleRegion.R

@@ -53,15 +53,19 @@ CloseBySingleRegion <- function(
   genome <- match.arg(genome)
   arrayType <- match.arg(arrayType)
   CpGsOrdered_df <- OrderCpGsByLocation(
-    CpGs_char, genome, arrayType, manifest_gr, output = "dataframe"
+    CpGs_char = CpGs_char,
+    genome = genome,
+    arrayType = arrayType,
+    manifest_gr = manifest_gr,
+    output = "dataframe"
   )
 
-  ### Find close by clusters ###
+  # Find close by clusters
   chr <- CpGsOrdered_df$chr
   pos <- CpGsOrdered_df$pos
-  CpGsOrdered_df$cluster <- clusterMaker(chr, pos, maxGap = maxGap)
+  CpGsOrdered_df$cluster <- clusterMaker(chr = chr, pos = pos, maxGap = maxGap)
 
-  ### Create list of vectors of CpGs in each cluster ###
+  # Create list of vectors of CpGs in each cluster
   CpGsRegion_ls <- split(CpGsOrdered_df$cpg, CpGsOrdered_df$cluster)
 
   ### Filter for clusters with number of CpGs >= minCpGs ###

R/CoMethSingleRegion.R

@@ -77,57 +77,66 @@ CoMethSingleRegion <- function(CpGs_char,
   genome <- match.arg(genome)
   method <- match.arg(method)
 
+  
   ### Order CpGs by genomic location ###
   CpGsOrdered_df <- OrderCpGsByLocation(
-    CpGs_char, genome, arrayType, manifest_gr, output = "dataframe"
+    CpGs_char = CpGs_char,
+    genome = genome,
+    arrayType = arrayType,
+    manifest_gr = manifest_gr,
+    output = "dataframe"
   )
 
+  
   ### Extract beta matrix for the input CpGs ###
   # take common cpgs in beta matrix and the region first
   commonCpGs_char <- intersect(CpGsOrdered_df$cpg, row.names(dnam))
 
-  if (length(commonCpGs_char) >= minCpGs){
-
-      betaCluster_mat <- dnam[commonCpGs_char, ]
-
-      ### Transpose beta matrix ###
-      betaClusterTransp_mat <- t(betaCluster_mat)
-
-      ### Mark comethylated CpGs ###
-      keepCpGs_df <- MarkComethylatedCpGs(
-        betaCluster_mat = betaClusterTransp_mat,
-        method = method,
-        betaToM = betaToM,
-        rDropThresh_num
-      )
-
-      ### Find contiguous comethylated regions ###
-      keepContiguousCpGs_df <- FindComethylatedRegions(
-        CpGs_df = keepCpGs_df
-      )
-
-      ### Split CpG dataframe by Subregion ###
-      keepContiguousCpGs_ls <- SplitCpGDFbyRegion(
-        keepContiguousCpGs_df, genome, arrayType, manifest_gr, returnAllCpGs
-      )
-
-      ### Create Output Data Frame  ###
-      coMethCpGs_df <- CreateOutputDF(
-        keepCpGs_df, keepContiguousCpGs_df, CpGsOrdered_df, returnAllCpGs
-      )
-
-      ### Create output list of data frame and CpGs by subregion ###
-      coMethCpGs_ls <- list(
-        contiguousRegions = coMethCpGs_df,
-        CpGsSubregions = keepContiguousCpGs_ls
-      )
-
-      coMethCpGs_ls
-
-  } else {
+  if (length(commonCpGs_char) < minCpGs){
     return(NULL)
+  } else {
+    
+    ###  Mark comethylated CpGs  ###
+    betaClusterT_mat <- t( dnam[commonCpGs_char, ] )
+    
+    keepCpGs_df <- MarkComethylatedCpGs(
+      betaCluster_mat = betaClusterT_mat,
+      method = method,
+      betaToM = betaToM,
+      rDropThresh_num
+    )
+    
+    # Find contiguous comethylated regions
+    keepContiguousCpGs_df <- FindComethylatedRegions(
+      CpGs_df = keepCpGs_df
+    )
+    
+    
+    ###  Split CpG dataframe by Subregion  ###
+    keepContiguousCpGs_ls <- SplitCpGDFbyRegion(
+      CpGsSubregions_df = keepContiguousCpGs_df,
+      genome = genome,
+      arrayType = arrayType,
+      manifest_gr = manifest_gr,
+      returnAllCpGs = returnAllCpGs
+    )
+    
+    
+    ###  Create Output  ###
+    coMethCpGs_df <- CreateOutputDF(
+      keepCpGs_df = keepCpGs_df,
+      keepContiguousCpGs_df = keepContiguousCpGs_df,
+      CpGsOrdered_df = CpGsOrdered_df,
+      returnAllCpGs = returnAllCpGs
+    )
+    
+    coMethCpGs_ls <- list(
+      contiguousRegions = coMethCpGs_df,
+      CpGsSubregions = keepContiguousCpGs_ls
+    )
+    
+    coMethCpGs_ls
+    
   }
 
-
-
 }

R/CpGsInfoAllRegions.R

@@ -18,6 +18,7 @@
 #' @param contPheno_char character string of the continuous phenotype to be
 #'    tested against methylation values
 #' @param covariates_char character vector of covariate variables names
+#' @param genome human genome of reference hg19 (default) or hg38
 #' @param arrayType Type of array, can be "450k" or "EPIC"
 #'
 #' @return a data frame with locations of the genomic region (Region), CpG ID
@@ -47,27 +48,47 @@
 CpGsInfoAllRegions <- function(AllRegionNames_char,
                                allRegions_gr = NULL,
                                betas_df, pheno_df,
-                               contPheno_char, covariates_char,
+                               contPheno_char,
+                               covariates_char,
+                               genome = c("hg19", "hg38"),
                                arrayType = c("450k", "EPIC")){
-
+  
+  ###  Inputs  ###
+  
+  # Available manifest files are: 
+  #   "EPIC.hg19.manifest"  "EPIC.hg38.manifest"
+  #   "HM450.hg19.manifest" "HM450.hg38.manifest"
+  genome <- match.arg(genome)
   arrayType <- match.arg(arrayType)
+  manifest <- paste(
+    switch(arrayType, "450k" = "HM450", "EPIC" = "EPIC"),
+    genome, "manifest",
+    sep = "."
+  )
+  CpGlocations.gr <- ImportSesameData(manifest)  
+  
+  # Regions
   if (!is.null(allRegions_gr)) {
     AllRegionNames_char <- as.character(allRegions_gr)
   }
 
+  
+  ###  Apply  ###
   resultsAllRegions_ls <- lapply(
     AllRegionNames_char,
     FUN = CpGsInfoOneRegion,
     region_gr = NULL,
-    betas_df,
-    pheno_df,
-    contPheno_char,
-    covariates_char,
-    arrayType
+    betas_df = betas_df,
+    pheno_df = pheno_df,
+    contPheno_char = contPheno_char,
+    covariates_char = covariates_char,
+    arrayType = arrayType,
+    manifest_gr = CpGlocations.gr
   )
 
+  
+  ###  Return  ###
   resultsAllRegions_df <- do.call(rbind, resultsAllRegions_ls)
-
   unique(resultsAllRegions_df)
 
 }

R/CpGsInfoOneRegion.R

@@ -14,7 +14,12 @@
 #' @param contPheno_char character string of the continuous phenotype to be
 #'   tested against methylation values
 #' @param covariates_char character vector of covariate variables names
+#' @param genome human genome of reference hg19 (default) or hg38
 #' @param arrayType Type of array, can be "450k" or "EPIC"
+#' @param manifest_gr A GRanges object with the genome manifest (as returned by
+#'   \code{\link[ExperimentHub]{ExperimentHub}} or by
+#'   \code{\link{ImportSesameData}}). This function by default ignores this
+#'   argument in favour of the \code{genome} and \code{arrayType} arguments.
 #'
 #' @return a data frame with location of the genomic region (Region), CpG ID
 #'   (cpg), chromosome (chr), position (pos), results for testing association of
@@ -55,10 +60,15 @@ CpGsInfoOneRegion <- function(
   pheno_df,
   contPheno_char,
   covariates_char = NULL,
-  arrayType = c("450k","EPIC")
+  genome = c("hg19", "hg38"),
+  arrayType = c("450k", "EPIC"),
+  manifest_gr = NULL
 ){
   # browser()
 
+  
+  ###  Inputs  ###
+  genome <- match.arg(genome)
   arrayType <- match.arg(arrayType)
 
   switch(
@@ -74,29 +84,39 @@ CpGsInfoOneRegion <- function(
 
   ### Extract individual CpGs in the region ###
   if (is.null(region_gr)) {
-    CpGsToTest_char <- GetCpGsInRegion(regionName_char, arrayType = arrayType)
+    
+    CpGsToTest_char <- GetCpGsInRegion(
+      regionName_char = regionName_char,
+      genome = genome,
+      arrayType = arrayType, 
+      manifest_gr = manifest_gr
+    )
+    
   } else {
+    
     CpGsToTest_char <- GetCpGsInRegion(
-      region_gr = region_gr, arrayType = arrayType
+      region_gr = region_gr,
+      genome = genome,
+      arrayType = arrayType, 
+      manifest_gr = manifest_gr
     )
     regionName_char <- as.character(region_gr)
+    
   }
 
-  ### Transpose dnam from wide to long ###
-  CpGsBeta_df <- betas_df[
-    which(rownames(betas_df) %in% CpGsToTest_char),
-  ]
-
-  ### Calculate M values ###
+  
+  ###  Wrangle and Tidy Data  ###
+  CpGsBeta_df <- betas_df[rownames(betas_df) %in% CpGsToTest_char, ]
   CpGsMvalue_df <- log2(CpGsBeta_df / (1 - CpGsBeta_df))
 
-  ### Match samples to test in pheno and beta data frames ###
+  # Match samples to test in pheno and beta data frames
   rownames(pheno_df) <- pheno_df$Sample
   samplesToTest <- intersect(colnames(CpGsMvalue_df), rownames(pheno_df))
   phenoTest_df <- pheno_df[samplesToTest, ]
-  CpGsMvalueTest_df <- CpGsMvalue_df[ ,samplesToTest]
+  CpGsMvalueTest_df <- CpGsMvalue_df[ , samplesToTest]
 
-  ### Run linear model for each CpG ###
+  
+  ###  Function to run linear model for each CpG  ###
   if (is.null(covariates_char)){
 
     lmF <- function(Mvalue) {
@@ -116,13 +136,23 @@ CpGsInfoOneRegion <- function(
 
   }
 
-  resultAllCpGs <- data.frame(t(apply(CpGsMvalueTest_df, 1, lmF)))
-
-  ### Return results ###
+  
+  ###  Run the Models  ###
+  resultAllCpGs <- data.frame(
+    t( apply(CpGsMvalueTest_df, 1, lmF) )
+  )
   colnames(resultAllCpGs) <- c("slopeEstimate", "slopePval")
+  
+
+  ###  Wrangle results  ###
   CpGsLocation <- OrderCpGsByLocation(
-    CpGs_char = CpGsToTest_char, arrayType = arrayType, output = "dataframe"
+    CpGs_char = CpGsToTest_char,
+    genome = genome,
+    arrayType = arrayType,
+    manifest_gr = manifest_gr,
+    output = "dataframe"
   )
+  
   outDF <- merge(
     CpGsLocation, resultAllCpGs,
     by.x = "cpg", by.y = "row.names", sort = FALSE
@@ -136,14 +166,15 @@ CpGsInfoOneRegion <- function(
 
   outDF$slopeEstimate <- round(outDF$slopeEstimate, 4)
 
-  ### Add annotations
+  
+  ###  Add Annotations and Return  ###
   CpGsAnno_df <- annotation_df[
     CpGsToTest_char,
     c("UCSC_RefGene_Name", "UCSC_RefGene_Accession", "UCSC_RefGene_Group")
   ]
 
   outAnno_df <- merge(
-    outDF,  CpGsAnno_df,
+    outDF, CpGsAnno_df,
     by.x = "cpg", by.y = "row.names", sort = FALSE
   )
 

R/GetCpGsInRegion.R

@@ -40,8 +40,8 @@ GetCpGsInRegion <- function(
   ignoreStrand = TRUE
 ){
 
-  arrayType <- match.arg(arrayType)
   genome <- match.arg(genome)
+  arrayType <- match.arg(arrayType)
 
 
   ###  The GRanges Object  ###
@@ -65,7 +65,7 @@ GetCpGsInRegion <- function(
   } else {
     gr <- region_gr
   }
-  CpGlocations.gr <- subsetByOverlaps(CpGlocations.gr, gr)
+  CpGlocations.gr <- subsetByOverlaps(x = CpGlocations.gr, ranges = gr)
 
   OrderCpGsByLocation(
     names(CpGlocations.gr),