Improvements to API: formatting, namespace, dependencies, sanity checks, output

DESCRIPTION

@@ -5,6 +5,12 @@ Authors@R: person("Sahil", "Shah", email = "[email protected]", role
 Description: What the package does (one paragraph).
 Depends:
     R (>= 3.2.1)
+Imports:
+    graphics,
+    igraph,
+    KEGGgraph,
+    limma,
+    pcaPP
 License: What license is it under?
 LazyData: true
-RoxygenNote: 5.0.0
+RoxygenNote: 6.0.1

NAMESPACE

@@ -1,2 +1,16 @@
 # Generated by roxygen2: do not edit by hand
 
+export(calcAllPairsDistances)
+export(calcCorMatrix)
+export(calcGeneTStats)
+export(geneNIDG)
+export(plotRadiusVS)
+importFrom(graphics,abline)
+importFrom(graphics,mtext)
+importFrom(graphics,par)
+importFrom(graphics,plot)
+importFrom(igraph,V)
+importFrom(igraph,shortest.paths)
+importFrom(limma,lmFit)
+importFrom(limma,treat)
+importFrom(pcaPP,cor.fk)

R/Observed.DecayDE.R

@@ -11,41 +11,42 @@
 #' differential expression of other genes j in the neighborhood is inversely 
 #' related to the distance d(i,j) of gene j from gene i
 #'
-#' @param distance.matrix A matrix of the distances on the global network
-#' @param gene.id The name of the gene to which GeneSurrounder is applied 
-#' @param genes.assayedETnetwork The names of the genes that are assayed and on the network
+#' @param distance_matrix A matrix of the distances on the global network
+#' @param gene_id The name of the gene to which GeneSurrounder is applied 
+#' @param genes_assayedETnetwork The names of the genes that are assayed and on the network
 #' @param diameter The diameter of the global network 
-#' @param geneStats.observed A vector of the observed differential expression
+#' @param geneStats_observed A vector of the observed differential expression
+#' @importFrom pcaPP cor.fk
 #'
-
-Observed.DecayDE <- function(distance.matrix,
-							 gene.id,
-							 genes.assayedETnetwork,
-							 diameter,
-							 geneStats.observed){
-
-
-
-	distances <- distance.matrix[gene.id,
-							 genes.assayedETnetwork]
-
-
-	observed.tau_b <- vapply(1:diameter,function(RADIUS){
-
-
-		# Excludes gene j with distances > 0 
-		igenes.distances <- distances[distances <= RADIUS
-									 & distances > 0]
-		igenes.names <- names(igenes.distances)
-
-
-		return(
-
-			cor.fk(abs(geneStats.observed[igenes.names]), igenes.distances)
-
-			)
-
-	},
-	numeric(1))
-
+Observed.DecayDE <- function(distance_matrix, gene_id, genes_assayedETnetwork,
+                             diameter, geneStats_observed) {
+  if (class(genes_assayedETnetwork) %in% c("character", "factor")) {
+    if (!all(genes_assayedETnetwork %in% colnames(distance_matrix))) {
+      bad_genes <- genes_assayedETnetwork[!(genes_assayedETnetwork %in% colnames(distance_matrix))]
+      bad_genes_msg <- paste(bad_genes, collapse = ", ")
+      bad_genes_num <- length(bad_genes)
+      stop(paste("There are ", bad_genes_num, " genes that are not in 'distance_matrix'.",
+                  "Here is a list of those genes not included:", bad_genes_msg, sep = "\n"))
+    }
+  } else if (class(genes_assayedETnetwork) %in% c("numeric", "integer")) {
+    genes_assayedETnetwork <- as.integer(genes_assayedETnetwork)
+    if (any(genes_assayedETnetwork <= 0 | genes_assayedETnwork > ncol(distance_matrix))) {
+      stop("At least one of the supplied column numbers was less than zero or greater than ",
+           "the total number of columns in 'distance_matrix'.")
+    }
+  }
+
+  if (gene_id %in% rownames(distance_matrix)) {
+    stop("The supplied 'gene_id', '", gene_id, "', is not found in 'distance_matrix'.")
+  }
+
+  distances <- distance_matrix[gene_id, genes_assayedETnetwork]
+
+  observed_tau_b <- vapply(1:diameter, function(RADIUS) {
+    # Excludes gene j with distances > 0 
+    igenes_distances <- distances[distances <= RADIUS & distances > 0]
+    igenes_names <- names(igenes_distances)
+    tau_b <- abs(geneStats_observed[igenes_names])
+    pcaPP::cor.fk(tau_b, igenes_distances)
+  }, numeric(1))
 }

R/Observed.SI.R

@@ -1,42 +1,20 @@
-
-
-
 #' Sphere of Influence Step 
 #'
 #' Sphere of Influence assesses if a 
 #' candidate gene i meets the first criterion by testing if gene i 
 #' is more strongly correlated with its network neighbors than with a 
 #' random set of genes
 #'
-#' @param gene.id The name of the gene to which GeneSurrounder is applied  
-#' @param distance.matrix A matrix of the distances on the global network
-#' @param cor.matrix A matrix of correlations between the expression of the genes
+#' @param gene_id The name of the gene to which GeneSurrounder is applied  
+#' @param distance_matrix A matrix of the distances on the global network
+#' @param cor_matrix A matrix of correlations between the expression of the genes
 #' @param diameter The diameter of the global network 
-#' @param genes.assayedETnetwork The names of the genes that are assayed and on the network
+#' @param genes_assayedETnetwork The names of the genes that are assayed and on the network
 #'
-
-
-
-Observed.SI <- function(gene.id,
-						distance.matrix,
-						cor.matrix,
-						diameter,
-						genes.assayedETnetwork){
-
-
-
-
-	# Vector of cor between j and all other genes on network (excluding j)
-	cor.with.j <- cor.matrix[gene.id,
-							 setdiff(genes.assayedETnetwork,gene.id)]
-
-	distances.to.j <- distance.matrix[gene.id,
-									  setdiff(genes.assayedETnetwork,gene.id)]
-
-
-	observed.cor <- SumAbsCor(cor.with.j,
-			  				  diameter,
-			  				  distances.to.j)
-
-
-}
+Observed.SI <- function(gene_id, distance_matrix, cor_matrix,
+                        diameter, genes_assayedETnetwork){
+  # Vector of cor between j and all other genes on network (excluding j)
+  cor_with_j <- cor_matrix[gene_id, setdiff(genes_assayedETnetwork, gene_id)]
+  distances_to_j <- distance_matrix[gene_id, setdiff(genes_assayedETnetwork, gene_id)]
+  observed_cor <- SumAbsCor(cor_with_j, diameter, distances_to_j)
+}

R/Resample.DecayDE.R

@@ -5,94 +5,59 @@
 # author: Sahil Shah <[email protected]>
 # ==========================================================================
 
-
-
 #' Decay of Differential Expression step
 #'
 #' Decay of Differential Expression tests whether the magnitude of 
 #' differential expression of other genes j in the neighborhood is inversely 
 #' related to the distance d(i,j) of gene j from gene i
 #'
-#' @param distance.matrix A matrix of the distances on the global network
-#' @param gene.id The name of the gene to which GeneSurrounder is applied 
-#' @param genes.assayedETnetwork The names of the genes that are assayed and on the network
+#' @param distance_matrix A matrix of the distances on the global network
+#' @param gene_id The name of the gene to which GeneSurrounder is applied 
+#' @param genes_assayedETnetwork The names of the genes that are assayed and on the network
 #' @param diameter The diameter of the global network 
-#' @param perm.geneStats.matrix A matrix of the resampled differential expression
+#' @param perm_geneStats_matrix A matrix of the resampled differential expression
 #' @param sizes The number of genes assayed and on the network in each neighborhood
+#' @importFrom pcaPP cor.fk
 #'
-
-
-Resample.DecayDE <- function(distance.matrix,
-							 gene.id,
-							 genes.assayedETnetwork,
-							 diameter,
-							 perm.geneStats.matrix,
-							 sizes){
-
-
-	distances <- distance.matrix[gene.id,
-							 genes.assayedETnetwork]
-
-
-
-
-
-	num.genes <- length(genes.assayedETnetwork) - 1
-	geneid.d <- which(sizes == num.genes)[1]
-
-
-	# null.tau b matrix of 1000 row 34 columns b/c stacks columns
-	null.tau_b <- vapply(1:geneid.d,function(RADIUS){
-
-
-		# Excludes gene j with distances > 0 
-		igenes.distances <- distances[distances <= RADIUS
-									  & distances > 0]
-		igenes.names <- names(igenes.distances)
-
-
-
-
-
-		null <- vapply(1:nrow(perm.geneStats.matrix),function(resample.index){
-
-			return ( cor.fk( abs(perm.geneStats.matrix[resample.index,igenes.names]) , 
-							igenes.distances) 
-
-				   )
-
-		},
-		numeric(1))
-
-		# X <- cbind(igenes.distances,
-		# 	  t( abs(geneStats$resampled[,igenes.names]) ) 
-		# 	  )
-
-		# Y <- cor.fk(X)
-
-
-	},
-	numeric(1000)) # numeric is length of one of the columns vapply stacks
-
-	# 2:genid.d means number of rows in null.tau_b is not geneid.d
-
-	null.tau_b <- cbind(null.tau_b,
-					matrix(
-						rep(null.tau_b[, geneid.d],diameter - geneid.d),
-						nrow = 1000)
-
-						)
-
-
-
-	# cf. ?vapply  
-	# If FUN.VALUE is not an array, the result is a matrix with 
-	# length(FUN.VALUE) rows and length(X) columns,
-
-	# return matrix distance by resample number a la Resample.SI
-	return(t(null.tau_b))
-
-
+Resample.DecayDE <- function(distance_matrix, gene_id, genes_assayedETnetwork,
+                             diameter, perm_geneStats_matrix, sizes) {
+
+  numResamples <- nrow(perm_geneStats_matrix)
+  distances <- distance_matrix[gene_id, genes_assayedETnetwork]
+
+  num_genes <- length(genes_assayedETnetwork) - 1
+  geneid_d <- which(sizes == num_genes)[1]
+
+  # null_tau b matrix of numResamples (default 1000) rows and
+  # length(distance_matrix) columns b/c stacks columns
+  null_tau_b <- vapply(1:geneid_d, function(RADIUS) {
+    # Excludes gene j with distances > 0 
+    igenes_distances <- distances[distances <= RADIUS & distances > 0]
+    igenes_names <- names(igenes_distances)
+
+    null <- vapply(1:numResamples, function(resample_index) {
+      tau_b <- abs(perm_geneStats_matrix[resample_index, igenes_names])
+      pcaPP::cor.fk(tau_b, igenes_distances)
+    }, numeric(1))
+
+    # X <- cbind(igenes_distances,
+    # t( abs(geneStats$resampled[,igenes_names]) ) 
+    # )
+
+    # Y <- cor.fk(X)
+  }, numeric(numResamples)) # numeric is length of one of the columns vapply stacks
+
+  # 2:genid_d means number of rows in null_tau_b is not geneid_d
+  null_tau_b <- cbind(null_tau_b, matrix(
+    rep(null_tau_b[, geneid_d], diameter - geneid_d),
+    nrow = numResamples)
+  )
+
+  # cf. ?vapply  
+  # If FUN.VALUE is not an array, the result is a matrix with 
+  # length(FUN.VALUE) rows and length(X) columns,
+  # return matrix distance by resample number a la Resample.SI
+  return(t(null_tau_b))
 }