# HG changeset patch
# User davidvanzessen
# Date 1427374421 14400
# Node ID 4c4149fa0bcbcbba64ec4b13cd1bbc9171fa8e0b
# Parent  3f4b4ef46c7fc62691d481542b0d398f555f9a12
Uploaded

diff -r 3f4b4ef46c7f -r 4c4149fa0bcb merge_and_filter.r
--- a/merge_and_filter.r	Tue Mar 17 09:44:25 2015 -0400
+++ b/merge_and_filter.r	Thu Mar 26 08:53:41 2015 -0400
@@ -15,8 +15,9 @@
 
 
 summ = summ[summ$Functionality != "No results",]
-tmp = summ[summ$chunk_hit_percentage >= 70 & summ$nt_hit_percentage >= 70,]
-unmatched = summ[summ$chunk_hit_percentage < 70 & summ$nt_hit_percentage < 70,]
+higher_than=(summ$chunk_hit_percentage >= 70 & summ$nt_hit_percentage >= 70)
+tmp = summ[higher_than,]
+unmatched = summ[!higher_than,]
 unmatched = unmatched[,c("Sequence.ID", "chunk_hit_percentage", "nt_hit_percentage", "start_locations", "best_match")]
 summ = tmp
 rm(tmp)
@@ -29,7 +30,6 @@
 result = merge(result, mutationstats[,!(names(mutationstats) %in% names(result)[-1])], by="Sequence.ID")
 result = merge(result, hotspots[,!(names(hotspots) %in% names(result)[-1])], by="Sequence.ID")
 
-
 cleanup_columns = c("FR1.IMGT.Nb.of.mutations", 
                     "CDR1.IMGT.Nb.of.mutations", 
                     "FR2.IMGT.Nb.of.mutations", 
@@ -54,5 +54,8 @@
 
 result = result[,!(names(result) %in% c("past"))]
 
+print(paste("Number of rows in result:", nrow(result)))
+print(paste("Number of rows in unmatched:", nrow(unmatched)))
+
 write.table(x=result, file=output, sep="\t",quote=F,row.names=F,col.names=T)
 write.table(x=unmatched, file=unmatchedfile, sep="\t",quote=F,row.names=F,col.names=T)