Mercurial > repos > davidvanzessen > mutation_analysis

comparison imgt_loader.r @ 50:8ba6afa1247a draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author davidvanzessen
date Thu, 28 Jan 2016 09:21:25 -0500
parents
children d4e72eeea640
comparison
equal deleted inserted replaced
49:5c6b9e99d576 50:8ba6afa1247a
1 args <- commandArgs(trailingOnly = TRUE)
2
3 summ.file = args[1]
4 aa.file = args[2]
5 junction.file = args[3]
6 out.file = args[4]
7
8 summ = read.table(summ.file, sep="\t", header=T, quote="", fill=T)
9 aa = read.table(aa.file, sep="\t", header=T, quote="", fill=T)
10 junction = read.table(junction.file, sep="\t", header=T, quote="", fill=T)
11
12 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')
13 old_sequence_columns=c('CDR1.IMGT','CDR2.IMGT','CDR3.IMGT')
14 old_junction_columns=c('JUNCTION')
15
16 added_summary_columns=c('Functionality','V.REGION.identity..','V.REGION.identity.nt','D.REGION.reading.frame','AA.JUNCTION','Functionality.comment','Sequence')
17 added_sequence_columns=c('FR1.IMGT','FR2.IMGT','FR3.IMGT','CDR3.IMGT','JUNCTION','J.REGION','FR4.IMGT')
18 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")
19
20 out=summ[,c("Sequence.ID","JUNCTION.frame","V.GENE.and.allele","D.GENE.and.allele","J.GENE.and.allele")]
21
22 out[,"CDR1.Seq"] = aa[,"CDR1.IMGT"]
23 out[,"CDR1.Length"] = summ[,"CDR1.IMGT.length"]
24
25 out[,"CDR2.Seq"] = aa[,"CDR2.IMGT"]
26 out[,"CDR2.Length"] = summ[,"CDR2.IMGT.length"]
27
28 out[,"CDR3.Seq"] = aa[,"CDR3.IMGT"]
29 out[,"CDR3.Length"] = summ[,"CDR3.IMGT.length"]
30
31 out[,"CDR3.Seq.DNA"] = junction[,"JUNCTION"]
32 out[,"CDR3.Length.DNA"] = nchar(as.character(junction[,"JUNCTION"]))
33 out[,"Strand"] = summ[,"Orientation"]
34 out[,"CDR3.Found.How"] = "a"
35
36 out[,added_summary_columns] = summ[,added_summary_columns]
37
38 out[,added_sequence_columns] = aa[,added_sequence_columns]
39
40 out[,added_junction_columns] = junction[,added_junction_columns]
41
42 out[,"Top V Gene"] = gsub(".* ", "", gsub("\\*.*", "", summ[,"V.GENE.and.allele"]))
43 out[,"Top D Gene"] = gsub(".* ", "", gsub("\\*.*", "", summ[,"D.GENE.and.allele"]))
44 out[,"Top J Gene"] = gsub(".* ", "", gsub("\\*.*", "", summ[,"J.GENE.and.allele"]))
45
46 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')]
47
48 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')
49
50 out[,"VDJ Frame"] = as.character(out[,"VDJ Frame"])
51
52 fltr = out[,"VDJ Frame"] == "in-frame"
53 if(any(fltr)){
54 out[fltr, "VDJ Frame"] = "In-frame"
55 }
56
57 fltr = out[,"VDJ Frame"] == "null"
58 if(any(fltr)){
59 out[fltr, "VDJ Frame"] = "Out-of-frame"
60 }
61
62 fltr = out[,"VDJ Frame"] == "out-of-frame"
63 if(any(fltr)){
64 out[fltr, "VDJ Frame"] = "Out-of-frame"
65 }
66
67 fltr = out[,"VDJ Frame"] == ""
68 if(any(fltr)){
69 out[fltr, "VDJ Frame"] = "Out-of-frame"
70 }
71
72 for(col in c('Top V Gene','Top D Gene','Top J Gene')){
73 out[,col] = as.character(out[,col])
74 fltr = out[,col] == ""
75 if(any(fltr)){
76 out[fltr,col] = "NA"
77 }
78 }
79
80 write.table(out, out.file, sep="\t", quote=F, row.names=F, col.names=T)