Mercurial > repos > davidvanzessen > mutation_analysis
view imgt_loader.r @ 69:7acdcd5c52ef draft
Uploaded
author | davidvanzessen |
---|---|
date | Fri, 22 Apr 2016 05:07:10 -0400 |
parents | d4e72eeea640 |
children | c2e1d5a93f98 |
line wrap: on
line source
args <- commandArgs(trailingOnly = TRUE) summ.file = args[1] aa.file = args[2] junction.file = args[3] out.file = args[4] summ = read.table(summ.file, sep="\t", header=T, quote="", fill=T) aa = read.table(aa.file, sep="\t", header=T, quote="", fill=T) junction = read.table(junction.file, sep="\t", header=T, quote="", fill=T) 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') added_summary_columns=c('Functionality','V.REGION.identity..','V.REGION.identity.nt','D.REGION.reading.frame','AA.JUNCTION','Functionality.comment','Sequence') added_sequence_columns=c('FR1.IMGT','FR2.IMGT','FR3.IMGT','CDR3.IMGT','JUNCTION','J.REGION','FR4.IMGT') added_junction_columns=c("P3.V.nt.nb",'N.REGION.nt.nb','N1.REGION.nt.nb',"P5.D.nt.nb","P3.D.nt.nb",'N2.REGION.nt.nb',"P5.J.nt.nb","X3.V.REGION.trimmed.nt.nb","X5.D.REGION.trimmed.nt.nb","X3.D.REGION.trimmed.nt.nb","X5.J.REGION.trimmed.nt.nb","N.REGION","N1.REGION","N2.REGION") out=summ[,c("Sequence.ID","JUNCTION.frame","V.GENE.and.allele","D.GENE.and.allele","J.GENE.and.allele")] out[,"CDR1.Seq"] = aa[,"CDR1.IMGT"] out[,"CDR1.Length"] = summ[,"CDR1.IMGT.length"] out[,"CDR2.Seq"] = aa[,"CDR2.IMGT"] out[,"CDR2.Length"] = summ[,"CDR2.IMGT.length"] out[,"CDR3.Seq"] = aa[,"CDR3.IMGT"] out[,"CDR3.Length"] = summ[,"CDR3.IMGT.length"] out[,"CDR3.Seq.DNA"] = junction[,"JUNCTION"] out[,"CDR3.Length.DNA"] = nchar(as.character(junction[,"JUNCTION"])) out[,"Strand"] = summ[,"Orientation"] out[,"CDR3.Found.How"] = "a" out[,added_summary_columns] = summ[,added_summary_columns] out[,added_sequence_columns] = aa[,added_sequence_columns] out[,added_junction_columns] = junction[,added_junction_columns] out[,"Top V Gene"] = gsub(".* ", "", gsub("\\*.*", "", summ[,"V.GENE.and.allele"])) out[,"Top D Gene"] = gsub(".* ", "", gsub("\\*.*", "", summ[,"D.GENE.and.allele"])) out[,"Top J Gene"] = gsub(".* ", "", gsub("\\*.*", "", summ[,"J.GENE.and.allele"])) out = out[,c('Sequence.ID','JUNCTION.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','P3.V.nt.nb','N.REGION.nt.nb','N1.REGION.nt.nb','P5.D.nt.nb','P3.D.nt.nb','N2.REGION.nt.nb','P5.J.nt.nb','X3.V.REGION.trimmed.nt.nb','X5.D.REGION.trimmed.nt.nb','X3.D.REGION.trimmed.nt.nb','X5.J.REGION.trimmed.nt.nb','N.REGION','N1.REGION','N2.REGION')] 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') out[,"VDJ Frame"] = as.character(out[,"VDJ Frame"]) fltr = out[,"VDJ Frame"] == "in-frame" if(any(fltr)){ out[fltr, "VDJ Frame"] = "In-frame" } fltr = out[,"VDJ Frame"] == "null" if(any(fltr)){ out[fltr, "VDJ Frame"] = "Out-of-frame" } fltr = out[,"VDJ Frame"] == "out-of-frame" if(any(fltr)){ out[fltr, "VDJ Frame"] = "Out-of-frame" } fltr = out[,"VDJ Frame"] == "" if(any(fltr)){ out[fltr, "VDJ Frame"] = "Out-of-frame" } for(col in c('Top V Gene','Top D Gene','Top J Gene')){ out[,col] = as.character(out[,col]) fltr = out[,col] == "" if(any(fltr)){ out[fltr,col] = "NA" } } write.table(out, out.file, sep="\t", quote=F, row.names=F, col.names=T) print("created naive file output")