From d1aedf2cb9eb1d3d10fa0ecf184ab6dddb85d91a Mon Sep 17 00:00:00 2001
From: Julie Laffy <julie.laffy@weizmann.ac.il>
Date: Tue, 26 Sep 2023 16:18:32 +0300
Subject: [PATCH] aggr_dup_genes add function and doc

---
 NAMESPACE          |  5 +++++
 R/aggr_dup_genes.R | 45 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 50 insertions(+)
 create mode 100644 R/aggr_dup_genes.R

diff --git a/NAMESPACE b/NAMESPACE
index b7d44ab..d25e109 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -8,6 +8,7 @@ export(Jaccard)
 export(Overlap)
 export(Unlist)
 export(add_to_dataframe)
+export(aggr_dup_genes)
 export(aggr_gene_expr)
 export(alias2ensembl)
 export(alias2entrez)
@@ -226,6 +227,10 @@ importFrom(igraph,cluster_walktrap)
 importFrom(irlba,prcomp_irlba)
 importFrom(limma,alias2Symbol)
 importFrom(magrittr,"%>%")
+importFrom(matrixStats,colMaxs)
+importFrom(matrixStats,colMeans2)
+importFrom(matrixStats,colMedians)
+importFrom(matrixStats,colSums2)
 importFrom(matrixStats,rowSds)
 importFrom(mixtools,normalmixEM)
 importFrom(msigdbr,msigdbr)
diff --git a/R/aggr_dup_genes.R b/R/aggr_dup_genes.R
new file mode 100644
index 0000000..e53db1e
--- /dev/null
+++ b/R/aggr_dup_genes.R
@@ -0,0 +1,45 @@
+### summary statistics for dealing with duplicated gene names in exprssion matrix
+
+#' @title Aggregate duplicate gene rows
+#' @description Aggregate duplicate gene rows. Duplicate rows are collapsed into one by taking the colSums, colMeans, colMaxs or colMedians.
+#' @param m matrix with genes as rows
+#' @param stat statistic to aggregate duplicate gene rows. Defaults to 'sum'. Default: c("sum", "mean", "max", "median")
+#' @return matrix with unique gene rows
+#' @seealso 
+#'  \code{\link[matrixStats]{rowSums2}},\code{\link[matrixStats]{rowMeans2}},\code{\link[matrixStats]{rowRanges}},\code{\link[matrixStats]{rowMedians}}
+#' @rdname aggr_dup_genes
+#' @export 
+#' @importFrom matrixStats colSums2 colMeans2 colMaxs colMedians
+aggr_dup_genes = function(m, stat=c('sum','mean','max','median'), na.rm=T) {
+  
+  aggr_dup_gene = function(gene) {
+    mgene = m[rownames(m)==gene,]
+    statFUN(mgene, na.rm=na.rm)
+  }
+  
+  statFUN = switch(match.arg(stat),
+                   sum = matrixStats::colSums2,
+                   mean = matrixStats::colMeans2,
+                   max = matrixStats::colMaxs,
+                   median = matrixStats::colMedians)
+  
+  one_col=FALSE
+  if (ncol(m)==1) {
+	  one_col=TRUE
+	  m = cbind(m, m)
+  }
+  # gene order, keeping only first instance of dup.genes
+  gene.ord = rownames(m)[!duplicated(rownames(m))]
+  # find dup genes
+  dup.genes = unique(rownames(m)[duplicated(rownames(m))])
+  # for each dup genes, apply summary statistic and return numeric vec
+  # rbind dup gene vectors
+  mNew = do.call(rbind, sapply(dup.genes, aggr_dup_gene, simplify = F))
+  m = m[!rownames(m) %in% dup.genes, ]
+  m = rbind(m, mNew)
+  #m = m[gene.ord,]
+  if(one_col) {
+	  m = m[,1,drop=F]
+  }
+  m
+}