diff imgt_loader/imgt_loader.r @ 40:106275b54470 draft

Uploaded

--- a/imgt_loader/imgt_loader.r	Mon Apr 24 09:39:59 2017 -0400
+++ b/imgt_loader/imgt_loader.r	Tue May 30 07:26:33 2017 -0400
@@ -1,4 +1,5 @@
 args <- commandArgs(trailingOnly = TRUE)
+options(show.error.locations = TRUE)
 
 summ.file = args[1]
 sequences.file = args[2]
@@ -7,12 +8,19 @@
 gapped.aa.file = args[5]
 out.file = args[6]
 
+print(summ.file)
+print(out.file)
+
 summ = read.table(summ.file, sep="\t", header=T, quote="", fill=T)
 sequences = read.table(sequences.file, sep="\t", header=T, quote="", fill=T)
 aa = read.table(aa.file, sep="\t", header=T, quote="", fill=T)
 gapped.aa = read.table(gapped.aa.file, sep="\t", header=T, quote="", fill=T)
 junction = read.table(junction.file, sep="\t", header=T, quote="", fill=T)
 
+print(paste("nrow(summ)", nrow(summ)))
+
+write.table(summ, out.file, sep="\t", quote=F, row.names=F, col.names=T)
+
 fix_column_names = function(df){
     if("V.DOMAIN.Functionality" %in% names(df)){
         names(df)[names(df) == "V.DOMAIN.Functionality"] = "Functionality"
@@ -31,6 +39,8 @@
 gapped.aa = fix_column_names(gapped.aa)
 junction = fix_column_names(junction)
 
+print(paste("nrow(summ)", nrow(summ)))
+
 old_summary_columns=c('Sequence.ID','JUNCTION.frame','V.GENE.and.allele','D.GENE.and.allele','J.GENE.and.allele','CDR1.IMGT.length','CDR2.IMGT.length','CDR3.IMGT.length','Orientation')
 old_sequence_columns=c('CDR1.IMGT','CDR2.IMGT','CDR3.IMGT')
 old_junction_columns=c('JUNCTION')
@@ -43,6 +53,9 @@
 
 out=summ[,c("Sequence.ID","JUNCTION.frame","V.GENE.and.allele","D.GENE.and.allele","J.GENE.and.allele")]
 
+print(paste("nrow(summ)", nrow(summ)))
+print(paste("nrow(aa)", nrow(aa)))
+
 out[,"CDR1.Seq"] = aa[,"CDR1.IMGT"]
 out[,"CDR1.Length"] = summ[,"CDR1.IMGT.length"]
 
@@ -78,24 +91,20 @@
 names(out) = c('ID','VDJ Frame','Top V Gene','Top D Gene','Top J Gene','CDR1 Seq','CDR1 Length','CDR2 Seq','CDR2 Length','CDR3 Seq','CDR3 Length','CDR3 Seq DNA','CDR3 Length DNA','Strand','CDR3 Found How','Functionality','V-REGION identity %','V-REGION identity nt','D-REGION reading frame','AA JUNCTION','Functionality comment','Sequence','FR1-IMGT','FR2-IMGT','FR3-IMGT','CDR3-IMGT','JUNCTION','J-REGION','FR4-IMGT','P3V-nt nb','N-REGION-nt nb','N1-REGION-nt nb','P5D-nt nb','P3D-nt nb','N2-REGION-nt nb','P5J-nt nb','3V-REGION trimmed-nt nb','5D-REGION trimmed-nt nb','3D-REGION trimmed-nt nb','5J-REGION trimmed-nt nb','N-REGION','N1-REGION','N2-REGION', 'P5.D1.nt.nb', 'P3.D1.nt.nb', 'N2.REGION.nt.nb', 'P5.D2.nt.nb', 'P3.D2.nt.nb', 'N3.REGION.nt.nb', 'P5.D3.nt.nb', 'P3.D2.nt.nb', 'N4.REGION.nt.nb', 'X5.D1.REGION.trimmed.nt.nb', 'X3.D1.REGION.trimmed.nt.nb', 'X5.D2.REGION.trimmed.nt.nb', 'X3.D2.REGION.trimmed.nt.nb', 'X5.D3.REGION.trimmed.nt.nb', 'X3.D3.REGION.trimmed.nt.nb', 'D.REGION.nt.nb', 'D1.REGION.nt.nb', 'D2.REGION.nt.nb', 'D3.REGION.nt.nb')
 
 out[,"VDJ Frame"] = as.character(out[,"VDJ Frame"])
-
 fltr = out[,"VDJ Frame"] == "in-frame"
-if(any(fltr)){
+if(any(fltr, na.rm=T)){
 	out[fltr, "VDJ Frame"] = "In-frame"
 }
-
 fltr = out[,"VDJ Frame"] == "null"
-if(any(fltr)){
+if(any(fltr, na.rm = T)){
 	out[fltr, "VDJ Frame"] = "Out-of-frame"
 }
-
 fltr = out[,"VDJ Frame"] == "out-of-frame"
-if(any(fltr)){
+if(any(fltr, na.rm = T)){
 	out[fltr, "VDJ Frame"] = "Out-of-frame"
 }
-
 fltr = out[,"VDJ Frame"] == ""
-if(any(fltr)){
+if(any(fltr, na.rm = T)){
 	out[fltr, "VDJ Frame"] = "Out-of-frame"
 }
 
@@ -103,7 +112,7 @@
 	out[,col] = as.character(out[,col])
 	fltr = out[,col] == ""
 	fltr[is.na(fltr)] = T
-	if(any(fltr)){
+	if(any(fltr, na.rm = T)){
 		out[fltr,col] = "NA"
 	}
 }