# HG changeset patch
# User davidvanzessen
# Date 1444313248 14400
# Node ID 7dd7cefcf72d33724cee0a4134fafc766382eab3
# Parent  7658e9f3d416d35f7a1f13c72a389cfa2c6d8298
Uploaded

diff -r 7658e9f3d416 -r 7dd7cefcf72d RScript.r
--- a/RScript.r	Thu Oct 08 05:46:16 2015 -0400
+++ b/RScript.r	Thu Oct 08 10:07:28 2015 -0400
@@ -190,6 +190,11 @@
       first.rows = patient.fuzzy[first.match.filter,]
       second.rows = patient.fuzzy[second.match.filter,]
 
+      first.rows.v = table(first.rows$V_Segment_Major_Gene)
+      first.rows.v = names(first.rows.v[which.max(first.rows.v)])
+      first.rows.j = table(first.rows$J_Segment_Major_Gene)
+      first.rows.j = names(first.rows.j[which.max(first.rows.j)])
+
       first.sum = data.frame(merge = first.clone.sequence,
                              Patient = patient,
                              Receptor = first.rows[1,"Receptor"],
@@ -199,8 +204,8 @@
                              Log10_Frequency = log10(sum(first.rows$Frequency)),
                              Total_Read_Count = sum(first.rows$Total_Read_Count),
                              dsPerM = sum(first.rows$dsPerM),
-                             J_Segment_Major_Gene = sort(table(first.rows$J_Segment_Major_Gene),decreasing=TRUE)[1],
-                             V_Segment_Major_Gene = sort(table(first.rows$V_Segment_Major_Gene),decreasing=TRUE)[1],
+                             J_Segment_Major_Gene = first.rows.j,
+                             V_Segment_Major_Gene = first.rows.v,
                              Clone_Sequence = first.clone.sequence,
                              CDR3_Sense_Sequence = first.rows[1,"CDR3_Sense_Sequence"],
                              Related_to_leukemia_clone = F,
@@ -213,6 +218,11 @@
                              min_cell_paste = first.rows[1,"min_cell_paste"])
 
       if(nrow(second.rows) > 0){
+        second.rows.v = table(second.rows$V_Segment_Major_Gene)
+        second.rows.v = names(second.rows.v[which.max(second.rows.v)])
+        second.rows.j = table(second.rows$J_Segment_Major_Gene)
+        second.rows.j = names(second.rows.j[which.max(second.rows.j)])
+
         second.sum = data.frame(merge = first.clone.sequence,
                                Patient = patient,
                                Receptor = second.rows[1,"Receptor"],
@@ -222,8 +232,8 @@
                                Log10_Frequency = log10(sum(second.rows$Frequency)),
                                Total_Read_Count = sum(second.rows$Total_Read_Count),
                                dsPerM = sum(second.rows$dsPerM),
-                               J_Segment_Major_Gene = sort(table(second.rows$J_Segment_Major_Gene),decreasing=TRUE)[1],
-                               V_Segment_Major_Gene = sort(table(second.rows$V_Segment_Major_Gene),decreasing=TRUE)[1],
+                               J_Segment_Major_Gene = second.rows.j,
+                               V_Segment_Major_Gene = second.rows.v,
                                Clone_Sequence = first.clone.sequence,
                                CDR3_Sense_Sequence = second.rows[1,"CDR3_Sense_Sequence"],
                                Related_to_leukemia_clone = F,
@@ -238,11 +248,13 @@
         patientMerge = rbind(patientMerge, merge(first.sum, second.sum, by="merge"))
         patient.fuzzy = patient.fuzzy[!(first.match.filter | second.match.filter),]
 
+        hidden.clone.sequences = c(first.rows[-1,"Clone_Sequence"], second.rows[second.rows$Clone_Sequence != first.clone.sequence,"Clone_Sequence"])
+        merge.list[["second"]] = append(merge.list[["second"]], hidden.clone.sequences)
 
         if(sum(first.match.filter) == 1 & sum(second.match.filter) == 1){
           second.clone.sequence = patient.fuzzy[second.match.filter, "Clone_Sequence"]
-          if(nchar(first.clone.sequence) == nchar(second.clone.sequence)){
-            merge.list[["second"]] = append(merge.list[["second"]], second.clone.sequence)
+          if(first.clone.sequence != second.clone.sequence){
+            #merge.list[["second"]] = append(merge.list[["second"]], second.clone.sequence)
           }
         }