annotate Build_tissue-specific_expression_dataset.R @ 1:cfcc7a780991 draft default tip

planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
author proteore
date Thu, 13 Dec 2018 03:59:41 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
1
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
2
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
3 select_HPAimmunohisto<-function(hpa_ref, tissue, level, reliability) {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
4 HPA.normal = read.table(hpa_ref,header=TRUE,sep="\t",stringsAsFactors = FALSE)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
5 if (tissue == "tissue") {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
6 tissue <- unique(HPA.normal$Tissue)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
7 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
8 if (level == "level") {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
9 level <- unique(HPA.normal$Level)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
10 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
11 if (reliability == "reliability") {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
12 reliability <- unique(HPA.normal$Reliability)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
13 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
14 res.imm <- subset(HPA.normal, Tissue%in%tissue & Level%in%level & Reliability%in%reliability)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
15 return(res.imm)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
16 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
17
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
18
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
19
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
20 select_HPARNAseq<-function(hpa_ref, sample) {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
21 HPA.rnaTissue = read.table(hpa_ref,header=TRUE,sep="\t",stringsAsFactors = FALSE)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
22 res.rna <- subset(HPA.rnaTissue, Sample%in%sample, select = -c(Unit))
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
23 colnames(res.rna)[which(colnames(res.rna) == 'Value')] <- 'Value (TPM unit)'
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
24 return(res.rna)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
25 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
26
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
27 main <- function() {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
28 args <- commandArgs(TRUE)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
29 if(length(args)<1) {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
30 args <- c("--help")
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
31 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
32
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
33 # Help section
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
34 if("--help" %in% args) {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
35 cat("Selection and Annotation HPA
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
36 Arguments:
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
37 --data_source: immuno/rnaseq
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
38 --hpe_ref: path to reference file normal_tissue.tsv/rna_tissue.tsv)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
39 if immuno:
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
40 --tissue: list of tissues
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
41 --level: Not detected, Low, Medium, High
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
42 --reliability: Supported, Approved, Enhanced, Uncertain
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
43 if rnaseq:
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
44 --sample: Sample tissues
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
45 --output: output filename \n")
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
46 q(save="no")
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
47 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
48
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
49 # Parse arguments
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
50 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
51 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
52 args <- as.list(as.character(argsDF$V2))
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
53 names(args) <- argsDF$V1
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
54
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
55 # Extract options
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
56 data_source = args$data_source
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
57 hpa_ref = args$hpa_ref
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
58 if (data_source == "immuno") {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
59 tissue = strsplit(args$tissue, ",")[[1]]
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
60 level = strsplit(args$level, ",")[[1]]
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
61 reliability = strsplit(args$reliability, ",")[[1]]
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
62 # Calculation
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
63 res = select_HPAimmunohisto(hpa_ref, tissue, level, reliability)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
64 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
65 else if (data_source == "rnaseq") {
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
66 sample = strsplit(args$sample, ",")[[1]]
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
67 # Calculation
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
68 res = select_HPARNAseq(hpa_ref, sample)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
69 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
70
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
71 # Write output
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
72 output = args$output
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
73 write.table(res, output, sep = "\t", quote = FALSE, row.names = FALSE)
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
74 }
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
75
cfcc7a780991 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
76 main()