report: mmseqs2: highlight only combis included in top N shown (issue #119)

workflow_report/scripts/utils.R

@@ -621,15 +621,18 @@ plot_diamondDB_density2d <- function(df){
     return(pp)
 }
 
-plot_mmseqs2_overlap <- function(df){
+plot_mmseqs2_overlap <- function(df, topN=20){
     # cluster summary: number of proteins in clusters grouped by represented tools
     upsetr_input <- UpSetR::fromExpression(t(df %>% select(members))[1,,drop=TRUE])
     testit::assert(all( colnames(upsetr_input) %in% ASM_TOOL_NAMES ))
     upsetr_input <- upsetr_input[,ASM_TOOL_NAMES,drop=FALSE]
+    topN_combis <- rownames(df[with(df, order(-members)), ])[1:topN]
     UpSetR::upset(
         upsetr_input,
+        keep.order=TRUE,
         nsets=ncol(upsetr_input),
-        nintersects=20,
+        sets=ASM_TOOL_NAMES,
+        nintersects=topN,
         order.by="freq",
         decreasing=TRUE,
         # titles
@@ -641,7 +644,7 @@ plot_mmseqs2_overlap <- function(df){
         queries=c(
             # unique proteins
             lapply(
-                colnames(upsetr_input),
+                intersect(colnames(upsetr_input), topN_combis),
                 function(x){ list(query=intersects, params=list(x), active=TRUE, color=MMSEQS2_UPSETR_HIGHLIGHTS["unique"]) }
             ),
             # all tools