diff sequence_overview.r @ 81:a778156dad3d draft

Uploaded
author davidvanzessen
date Tue, 17 May 2016 04:17:38 -0400
parents 0513b46178c4
children 564c4f6da203
line wrap: on
line diff
--- a/sequence_overview.r	Thu May 12 10:51:55 2016 -0400
+++ b/sequence_overview.r	Tue May 17 04:17:38 2016 -0400
@@ -5,6 +5,8 @@
 gene.matches = args[1]
 sequence.file = args[2]
 outputdir = args[3]
+NToverview.file = paste(outputdir, "ntoverview.txt", sep="/")
+NTsum.file = paste(outputdir, "ntsum.txt", sep="/")
 main.html = "index.html"
 
 setwd(outputdir)
@@ -21,11 +23,11 @@
 
 #dat = data.frame(data.table(dat)[, list(freq=.N), by=c("best_match", "seq_conc")])
 
-dat = data.frame(table(dat$best_match, dat$seq_conc, dat$Functionality))
+dat = data.frame(table(dat$seq_conc))
 
 dat = dat[dat$Freq > 1,]
 
-names(dat) = c("best_match", "seq_conc", "Functionality", "Freq")
+names(dat) = c("seq_conc", "Freq")
 
 dat$seq_conc = factor(dat$seq_conc)
 
@@ -100,3 +102,58 @@
 }
 
 cat("</table>", file=main.html, append=T)
+
+
+#ACGT overview
+
+
+
+NToverview = genes[,c("Sequence.ID", "best_match")]
+sequences$seq = paste(sequences$CDR2.IMGT, sequences$CDR2.IMGT, sequences$FR2.IMGT, sequences$FR3.IMGT, sep="_")
+
+NToverview = merge(NToverview, sequences[,c("Sequence.ID", "seq")], by="Sequence.ID")
+
+NToverview$A = nchar(gsub("[^Aa]", "", NToverview$seq))
+NToverview$C = nchar(gsub("[^Cc]", "", NToverview$seq))
+NToverview$G = nchar(gsub("[^Gg]", "", NToverview$seq))
+NToverview$T = nchar(gsub("[^Tt]", "", NToverview$seq))
+
+NTsum = data.frame(Sequence.ID="-", best_match="Sum", seq="-", A = sum(NToverview$A), C = sum(NToverview$C), G = sum(NToverview$G), T = sum(NToverview$T))
+
+print(names(NToverview))
+print(names(NTsum))
+
+NToverview = rbind(NToverview, NTsum)
+
+write.table(NToverview, NToverview.file, quote=F, sep="\t", row.names=F, col.names=T)
+#write.table(NTsum, NTsum.file, quote=F, sep="\t", row.names=F, col.names=T)
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+