annotate sequence_overview.r @ 78:b523ce95d857 draft

Uploaded
author davidvanzessen
date Wed, 11 May 2016 10:29:33 -0400
parents c5c86d15cb94
children 0513b46178c4
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
76
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
1 library(reshape2)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
2
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
3 args <- commandArgs(trailingOnly = TRUE)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
4
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
5 gene.matches = args[1]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
6 sequence.file = args[2]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
7 outputdir = args[3]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
8 main.html = "index.html"
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
9
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
10 setwd(outputdir)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
11
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
12 genes = read.table(gene.matches, header=T, sep="\t", fill=T)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
13 sequences = read.table(sequence.file, header=T, sep="\t", fill=T, stringsAsFactors=F, quote="")
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
14
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
15 dat = merge(sequences, genes, by="Sequence.ID")
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
16
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
17 dat$seq_conc = paste(dat$CDR1.IMGT, dat$CDR2.IMGT, dat$CDR3.IMGT, dat$FR2.IMGT, dat$FR3.IMGT)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
18
78
b523ce95d857 Uploaded
davidvanzessen
parents: 77
diff changeset
19 IDs = dat[,c("Sequence.ID", "seq_conc", "best_match", "Functionality")]
76
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
20 IDs$best_match = as.character(IDs$best_match)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
21
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
22 #dat = data.frame(data.table(dat)[, list(freq=.N), by=c("best_match", "seq_conc")])
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
23
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
24 dat = data.frame(table(dat$best_match, dat$seq_conc))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
25
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
26 dat = dat[dat$Freq > 1,]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
27
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
28 names(dat) = c("best_match", "seq_conc", "Freq")
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
29
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
30 dat$seq_conc = factor(dat$seq_conc)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
31
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
32 dat = dat[order(nchar(as.character(dat$seq_conc))),]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
33
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
34 #writing html from R...
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
35 td = function(val) { paste("<td>", val, "</td>", sep="") }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
36 tr = function(val) { capture.output(cat("<tr>", td(val), "</tr>", sep="")) }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
37 make.link = function(id, clss, val) { paste("<a href='", clss, "_", id, ".html'>", val, "</a>", sep="") }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
38 tbl = function(df) { res = "<table border='1'>"; for(i in 1:nrow(df)){ res = paste(res, tr(df[i,]), sep=""); }; res = paste(res, "</table>"); }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
39
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
40 cat("<table border='1'>", file=main.html, append=F)
77
c5c86d15cb94 Uploaded
davidvanzessen
parents: 76
diff changeset
41 cat("<caption>CDR1+CDR2+CDR3+FR2+FR3 sequences that show up more than once</caption>", file=main.html, append=T)
76
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
42 cat("<tr><th>Sequence</th><th>ca1</th><th>ca2</th><th>cg1</th><th>cg2</th><th>cg3</th><th>cg4</th><th>cm</th></tr>", file=main.html, append=T)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
43
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
44 for(i in 1:nrow(dat)){
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
45 ca1 = IDs[IDs$seq_conc == dat[i,c("seq_conc")] & IDs$best_match == "ca1",]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
46 ca2 = IDs[IDs$seq_conc == dat[i,c("seq_conc")] & IDs$best_match == "ca2",]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
47
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
48 cg1 = IDs[IDs$seq_conc == dat[i,c("seq_conc")] & IDs$best_match == "cg1",]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
49 cg2 = IDs[IDs$seq_conc == dat[i,c("seq_conc")] & IDs$best_match == "cg2",]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
50 cg3 = IDs[IDs$seq_conc == dat[i,c("seq_conc")] & IDs$best_match == "cg3",]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
51 cg4 = IDs[IDs$seq_conc == dat[i,c("seq_conc")] & IDs$best_match == "cg4",]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
52
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
53 cm = IDs[IDs$seq_conc == dat[i,c("seq_conc")] & IDs$best_match == "cm",]
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
54
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
55 id = as.numeric(dat[i,"seq_conc"])
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
56
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
57 if(nrow(ca1) > 0){
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
58 cat(tbl(ca1), file=paste("ca1_", id, ".html", sep=""))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
59 }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
60
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
61 if(nrow(ca2) > 0){
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
62 cat(tbl(ca2), file=paste("ca2_", id, ".html", sep=""))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
63 }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
64
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
65 if(nrow(cg1) > 0){
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
66 cat(tbl(cg1), file=paste("cg1_", id, ".html", sep=""))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
67 }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
68
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
69 if(nrow(cg2) > 0){
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
70 cat(tbl(cg2), file=paste("cg2_", id, ".html", sep=""))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
71 }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
72
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
73 if(nrow(cg3) > 0){
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
74 cat(tbl(cg3), file=paste("cg3_", id, ".html", sep=""))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
75 }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
76
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
77 if(nrow(cg4) > 0){
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
78 cat(tbl(cg4), file=paste("cg4_", id, ".html", sep=""))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
79 }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
80
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
81 if(nrow(cm) > 0){
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
82 cat(tbl(cm), file=paste("cm_", id, ".html", sep=""))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
83 }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
84
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
85 ca1.html = make.link(id, "ca1", nrow(ca1))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
86 ca2.html = make.link(id, "ca2", nrow(ca2))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
87
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
88 cg1.html = make.link(id, "cg1", nrow(cg1))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
89 cg2.html = make.link(id, "cg2", nrow(cg2))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
90 cg3.html = make.link(id, "cg3", nrow(cg3))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
91 cg4.html = make.link(id, "cg4", nrow(cg4))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
92
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
93 cm.html = make.link(id, "cm", nrow(cm))
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
94
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
95 rw = c(as.character(dat[i,"seq_conc"]), ca1.html, ca2.html, cg1.html, cg2.html, cg3.html, cg4.html, cm.html)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
96
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
97 cat(tr(rw), file=main.html, append=T)
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
98 }
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
99
becea91089ed Uploaded
davidvanzessen
parents:
diff changeset
100 cat("</table>", file=main.html, append=T)