From 77c7bcb36680d83b41890640bcd12ae87bc31df1 Mon Sep 17 00:00:00 2001
From: Vlad Petyuk <petyuk@gmail.com>
Date: Mon, 6 May 2024 12:16:29 -0700
Subject: [PATCH] resolving gene-site ambiguity

---
 R/read_FragPipe_TMT.R | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/R/read_FragPipe_TMT.R b/R/read_FragPipe_TMT.R
index 017ed2b..609292b 100644
--- a/R/read_FragPipe_TMT.R
+++ b/R/read_FragPipe_TMT.R
@@ -59,12 +59,21 @@ read_FragPipe_TMT <- function(path = NULL, org_to_retain = NULL, use_gene_as_pro
   else if (grepl("single-site", basename(path_to_file))) {
      if(use_gene_as_prot_id){
         df <- df %>%
+           filter(Gene != "") %>%
            mutate(rowname = paste0(Gene,
                                    "-",
                                    sub("[^_]*_([A-Z]\\d+)","\\1",Index)))
         if(anyDuplicated(df$rowname)){
-           stop("Duplicates in the gene-based site names. Switch to use_gene_as_prot_id = FALSE.")
-        }
+           # let's try to resolve by ReferenceIntensity
+           if(!("ReferenceIntensity" %in% colnames(df))){
+              stop("Duplicates in the gene-based site names. Can't resolve ambiguity.
+                   Switch to use_gene_as_prot_id = FALSE.")
+           } else {
+              df <- df %>%
+                 group_by(rowname) %>%
+                 slice_max(ReferenceIntensity)
+          }
+       }
      }else{
         df <- df %>% mutate(rowname = Index)
      }