# HG changeset patch
# User davidvanzessen
# Date 1424787148 18000
# Node ID 6904186d13b9272424114628c6f1661be9d945b2
# Parent  0940835d259c06c0aecdcd2e26421d5cbd2c705f
Uploaded

diff -r 0940835d259c -r 6904186d13b9 RScript.r
--- a/RScript.r	Tue Feb 24 07:30:36 2015 -0500
+++ b/RScript.r	Tue Feb 24 09:12:28 2015 -0500
@@ -265,7 +265,7 @@
   patientMerge23 = merge(patient2, patient3, by="merge")
   patientMerge23$thresholdValue = pmax(patientMerge23[,onx], patientMerge23[,ony])
   
-  scatterplot_data_columns = c("Clone_Sequence", "Frequency", "normalized_read_count")
+  scatterplot_data_columns = c("Clone_Sequence", "Frequency", "normalized_read_count", "V_Segment_Major_Gene", "J_Segment_Major_Gene")
   scatterplot_data = rbind(patient1[,scatterplot_data_columns], patient2[,scatterplot_data_columns], patient3[,scatterplot_data_columns])
   scatterplot_data = scatterplot_data[!duplicated(scatterplot_data$Clone_Sequence),]
   scatterplot_data$type = factor(x="single", levels=c("In one", "In two", "In three"))
@@ -290,13 +290,14 @@
     J_Segment = paste(".*", as.character(product[iter,J_SegmentIndex]), ".*", sep="")
     #all = (grepl(V_Segment, patientMerge$V_Segment_Major_Gene.x) & grepl(J_Segment, patientMerge$J_Segment_Major_Gene.x) & patientMerge[,onx] > threshhold & patientMerge[,ony] > threshhold & patientMerge[,onz] > threshhold) 
     all = (grepl(V_Segment, patientMerge$V_Segment_Major_Gene.x) & grepl(J_Segment, patientMerge$J_Segment_Major_Gene.x) & patientMerge$thresholdValue > threshhold)
-    one = (grepl(V_Segment, patient1$V_Segment_Major_Gene) & grepl(J_Segment, patient1$J_Segment_Major_Gene) & patient1[,on] > threshhold & !(patient1$Clone_Sequence %in% patientMerge[all,]$merge))
-    two = (grepl(V_Segment, patient2$V_Segment_Major_Gene) & grepl(J_Segment, patient2$J_Segment_Major_Gene) & patient2[,on] > threshhold & !(patient2$Clone_Sequence %in% patientMerge[all,]$merge))
-    three = (grepl(V_Segment, patient3$V_Segment_Major_Gene) & grepl(J_Segment, patient3$J_Segment_Major_Gene) & patient3[,on] > threshhold & !(patient3$Clone_Sequence %in% patientMerge[all,]$merge))
     
-    one_two = (grepl(V_Segment, patientMerge12$V_Segment_Major_Gene.x) & grepl(J_Segment, patientMerge12$J_Segment_Major_Gene.x) & patientMerge12$thresholdValue > threshhold)
-    one_three = (grepl(V_Segment, patientMerge13$V_Segment_Major_Gene.x) & grepl(J_Segment, patientMerge13$J_Segment_Major_Gene.x) & patientMerge13$thresholdValue > threshhold)
-    two_three = (grepl(V_Segment, patientMerge23$V_Segment_Major_Gene.x) & grepl(J_Segment, patientMerge23$J_Segment_Major_Gene.x) & patientMerge23$thresholdValue > threshhold)
+    one_two = (grepl(V_Segment, patientMerge12$V_Segment_Major_Gene.x) & grepl(J_Segment, patientMerge12$J_Segment_Major_Gene.x) & patientMerge12$thresholdValue > threshhold & !(patientMerge12$Clone_Sequence %in% patientMerge[all,]$merge))
+    one_three = (grepl(V_Segment, patientMerge13$V_Segment_Major_Gene.x) & grepl(J_Segment, patientMerge13$J_Segment_Major_Gene.x) & patientMerge13$thresholdValue > threshhold & !(patientMerge13$Clone_Sequence %in% patientMerge[all,]$merge))
+    two_three = (grepl(V_Segment, patientMerge23$V_Segment_Major_Gene.x) & grepl(J_Segment, patientMerge23$J_Segment_Major_Gene.x) & patientMerge23$thresholdValue > threshhold & !(patientMerge23$Clone_Sequence %in% patientMerge[all,]$merge))
+    
+    one = (grepl(V_Segment, patient1$V_Segment_Major_Gene) & grepl(J_Segment, patient1$J_Segment_Major_Gene) & patient1[,on] > threshhold & !(patient1$Clone_Sequence %in% patientMerge[all,]$merge) & !(patientMerge12$Clone_Sequence %in% patientMerge[all,]$merge) & !(patientMerge13$Clone_Sequence %in% patientMerge[all,]$merge))
+    two = (grepl(V_Segment, patient2$V_Segment_Major_Gene) & grepl(J_Segment, patient2$J_Segment_Major_Gene) & patient2[,on] > threshhold & !(patient2$Clone_Sequence %in% patientMerge[all,]$merge) & !(patientMerge12$Clone_Sequence %in% patientMerge[all,]$merge) & !(patientMerge23$Clone_Sequence %in% patientMerge[all,]$merge))
+    three = (grepl(V_Segment, patient3$V_Segment_Major_Gene) & grepl(J_Segment, patient3$J_Segment_Major_Gene) & patient3[,on] > threshhold & !(patient3$Clone_Sequence %in% patientMerge[all,]$merge) & !(patientMerge13$Clone_Sequence %in% patientMerge[all,]$merge) & !(patientMerge23$Clone_Sequence %in% patientMerge[all,]$merge))
     
     read1Count = append(read1Count, sum(patient1[one,]$normalized_read_count) + sum(patientMerge[all,]$normalized_read_count.x))
     read2Count = append(read2Count, sum(patient2[two,]$normalized_read_count) + sum(patientMerge[all,]$normalized_read_count.y))
@@ -347,7 +348,7 @@
         write.table(dfTwo_three, file=paste(filenameTwo_three, ".txt", sep=""), quote=F, sep="\t", dec=",", row.names=F, col.names=T)
       }
     } else { #scatterplot data
-      scatterplot_locus_data = scatterplot_data
+      scatterplot_locus_data = scatterplot_data[grepl(V_Segment, scatterplot_data$V_Segment_Major_Gene) & grepl(J_Segment, scatterplot_data$J_Segment_Major_Gene),]
       scatterplot_locus_data$type = ifelse(scatterplot_locus_data$Clone_Sequence %in% patientMerge12[one_two,]$Clone_Sequence.x, "In two", "In one")
       scatterplot_locus_data$type = ifelse(scatterplot_locus_data$Clone_Sequence %in% patientMerge13[one_three,]$Clone_Sequence.x, "In two", "In one")
       scatterplot_locus_data$type = ifelse(scatterplot_locus_data$Clone_Sequence %in% patientMerge23[two_three,]$Clone_Sequence.x, "In two", "In one")