Mercurial > repos > shians > shrnaseq
annotate hairpinTool.R @ 8:548802b3492f
Version 1.0.9
- Added R session info to output
- Added email for bug reports
- Changed edgeR requirement to be stopping condition
- Changed names of gene-wise comparison tables to "gene_level" from "roast"
- Fixed selection of multiple genes for barcode plotting
| author | shian_su <registertonysu@gmail.com> |
|---|---|
| date | Fri, 02 May 2014 17:22:24 +1000 |
| parents | 91e411fcdecc |
| children | f1076bfb0ed1 |
| rev | line source |
|---|---|
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
1 # ARGS: 1.inputType -String specifying format of input (fastq or table) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
2 # IF inputType is "fastQ": |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
3 # 2*.fastqPath -One or more strings specifying path to fastq files |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
4 # 2.annoPath -String specifying path to hairpin annotation table |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
5 # 3.samplePath -String specifying path to sample annotation table |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
6 # 4.barStart -Integer specifying starting position of barcode |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
7 # 5.barEnd -Integer specifying ending position of barcode |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
8 # 6.hpStart -Integer specifying startins position of hairpin |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
9 # unique region |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
10 # 7.hpEnd -Integer specifying ending position of hairpin |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
11 # unique region |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
12 # ### |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
13 # IF inputType is "counts": |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
14 # 2.countPath -String specifying path to count table |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
15 # 3.annoPath -String specifying path to hairpin annotation table |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
16 # 4.samplePath -String specifying path to sample annotation table |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
17 # ### |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
18 # 8.cpmReq -Float specifying cpm requirement |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
19 # 9.sampleReq -Integer specifying cpm requirement |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
20 # 10.fdrThresh -Float specifying the FDR requirement |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
21 # 11.lfcThresh -Float specifying the log-fold-change requirement |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
22 # 12.workMode -String specifying exact test or GLM usage |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
23 # 13.htmlPath -String specifying path to HTML file |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
24 # 14.folderPath -STring specifying path to folder for output |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
25 # IF workMode is "classic" (exact test) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
26 # 15.pairData[2] -String specifying first group for exact test |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
27 # 16.pairData[1] -String specifying second group for exact test |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
28 # ### |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
29 # IF workMode is "glm" |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
30 # 15.contrastData -String specifying contrasts to be made |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
31 # 16.roastOpt -String specifying usage of gene-wise tests |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
32 # 17.hairpinReq -String specifying hairpin requirement for gene- |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
33 # wise test |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
34 # 18.selectOpt -String specifying type of selection for barcode |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
35 # plots |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
36 # 19.selectVals -String specifying members selected for barcode |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
37 # plots |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
38 # |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
39 # OUT: Bar Plot of Counts Per Index |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
40 # Bar Plot of Counts Per Hairpin |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
41 # MDS Plot |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
42 # Smear Plot |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
43 # Barcode Plots (If Genewise testing was selected) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
44 # Top Expression Table |
| 7 | 45 # Feature Counts Table |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
46 # HTML file linking to the ouputs |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
47 # |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
48 # Author: Shian Su - registertonysu@gmail.com - Jan 2014 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
49 |
| 2 | 50 # Record starting time |
| 51 timeStart <- as.character(Sys.time()) | |
| 52 | |
| 53 # Loading and checking required packages | |
| 54 library(methods, quietly=TRUE, warn.conflicts=FALSE) | |
| 55 library(statmod, quietly=TRUE, warn.conflicts=FALSE) | |
| 56 library(splines, quietly=TRUE, warn.conflicts=FALSE) | |
| 57 library(edgeR, quietly=TRUE, warn.conflicts=FALSE) | |
| 58 library(limma, quietly=TRUE, warn.conflicts=FALSE) | |
| 59 | |
| 60 if (packageVersion("edgeR") < "3.5.23") { | |
| 8 | 61 stop("Please update 'edgeR' to version >= 3.5.23 to run this tool") |
| 62 } | |
| 63 | |
| 64 if (packageVersion("limma")<"3.19.19") { | |
| 65 message("Update 'limma' to version >= 3.19.19 to see updated barcode graphs") | |
| 2 | 66 } |
| 67 | |
| 68 ################################################################################ | |
| 69 ### Function declarations | |
| 70 ################################################################################ | |
| 71 | |
| 7 | 72 # Function to load libaries without messages |
| 73 silentLibrary <- function(...) { | |
| 74 list <- c(...) | |
| 75 for (package in list){ | |
| 76 suppressPackageStartupMessages(library(package, character.only=TRUE)) | |
| 77 } | |
| 78 } | |
| 79 | |
| 2 | 80 # Function to sanitise contrast equations so there are no whitespaces |
| 81 # surrounding the arithmetic operators, leading or trailing whitespace | |
| 82 sanitiseEquation <- function(equation) { | |
| 83 equation <- gsub(" *[+] *", "+", equation) | |
| 84 equation <- gsub(" *[-] *", "-", equation) | |
| 85 equation <- gsub(" *[/] *", "/", equation) | |
| 86 equation <- gsub(" *[*] *", "*", equation) | |
| 87 equation <- gsub("^\\s+|\\s+$", "", equation) | |
| 88 return(equation) | |
| 89 } | |
| 90 | |
| 91 # Function to sanitise group information | |
| 92 sanitiseGroups <- function(string) { | |
| 93 string <- gsub(" *[,] *", ",", string) | |
| 94 string <- gsub("^\\s+|\\s+$", "", string) | |
| 95 return(string) | |
| 96 } | |
| 97 | |
| 98 # Function to change periods to whitespace in a string | |
| 99 unmake.names <- function(string) { | |
| 100 string <- gsub(".", " ", string, fixed=TRUE) | |
| 101 return(string) | |
| 102 } | |
| 103 | |
| 104 # Function has string input and generates an output path string | |
| 105 makeOut <- function(filename) { | |
| 106 return(paste0(folderPath, "/", filename)) | |
| 107 } | |
| 108 | |
| 109 # Function has string input and generates both a pdf and png output strings | |
| 110 imgOut <- function(filename) { | |
| 111 assign(paste0(filename, "Png"), makeOut(paste0(filename,".png")), | |
| 7 | 112 envir=.GlobalEnv) |
| 2 | 113 assign(paste0(filename, "Pdf"), makeOut(paste0(filename,".pdf")), |
| 7 | 114 envir=.GlobalEnv) |
| 2 | 115 } |
| 116 | |
| 117 # Create cat function default path set, default seperator empty and appending | |
| 118 # true by default (Ripped straight from the cat function with altered argument | |
| 119 # defaults) | |
| 7 | 120 cata <- function(..., file=htmlPath, sep="", fill=FALSE, labels=NULL, |
| 121 append=TRUE) { | |
| 2 | 122 if (is.character(file)) |
| 123 if (file == "") | |
| 124 file <- stdout() | |
| 125 else if (substring(file, 1L, 1L) == "|") { | |
| 126 file <- pipe(substring(file, 2L), "w") | |
| 127 on.exit(close(file)) | |
| 128 } | |
| 129 else { | |
| 130 file <- file(file, ifelse(append, "a", "w")) | |
| 131 on.exit(close(file)) | |
| 132 } | |
| 133 .Internal(cat(list(...), file, sep, fill, labels, append)) | |
| 134 } | |
| 135 | |
| 136 # Function to write code for html head and title | |
| 137 HtmlHead <- function(title) { | |
| 138 cata("<head>\n") | |
| 139 cata("<title>", title, "</title>\n") | |
| 140 cata("</head>\n") | |
| 141 } | |
| 142 | |
| 143 # Function to write code for html links | |
| 144 HtmlLink <- function(address, label=address) { | |
| 145 cata("<a href=\"", address, "\" target=\"_blank\">", label, "</a><br />\n") | |
| 146 } | |
| 147 | |
| 148 # Function to write code for html images | |
| 149 HtmlImage <- function(source, label=source, height=600, width=600) { | |
| 150 cata("<img src=\"", source, "\" alt=\"", label, "\" height=\"", height) | |
| 151 cata("\" width=\"", width, "\"/>\n") | |
| 152 } | |
| 153 | |
| 154 # Function to write code for html list items | |
| 155 ListItem <- function(...) { | |
| 156 cata("<li>", ..., "</li>\n") | |
| 157 } | |
| 158 | |
| 159 TableItem <- function(...) { | |
| 160 cata("<td>", ..., "</td>\n") | |
| 161 } | |
| 162 | |
| 163 TableHeadItem <- function(...) { | |
| 164 cata("<th>", ..., "</th>\n") | |
| 165 } | |
| 166 ################################################################################ | |
| 167 ### Input Processing | |
| 168 ################################################################################ | |
| 169 | |
| 170 # Grabbing arguments from command line | |
| 171 argv <- commandArgs(TRUE) | |
| 172 | |
| 173 # Remove fastq file paths after collecting from argument vector | |
| 174 inputType <- as.character(argv[1]) | |
| 175 if (inputType=="fastq") { | |
| 176 fastqPath <- as.character(gsub("fastq::", "", argv[grepl("fastq::", argv)], | |
| 177 fixed=TRUE)) | |
| 178 argv <- argv[!grepl("fastq::", argv, fixed=TRUE)] | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
179 annoPath <- as.character(argv[2]) |
| 2 | 180 samplePath <- as.character(argv[3]) |
| 181 barStart <- as.numeric(argv[4]) | |
| 182 barEnd <- as.numeric(argv[5]) | |
| 183 hpStart <- as.numeric(argv[6]) | |
| 184 hpEnd <- as.numeric(argv[7]) | |
| 185 } else if (inputType=="counts") { | |
| 186 countPath <- as.character(argv[2]) | |
| 187 annoPath <- as.character(argv[3]) | |
| 188 samplePath <- as.character(argv[4]) | |
| 189 } | |
| 190 | |
| 191 cpmReq <- as.numeric(argv[8]) | |
| 192 sampleReq <- as.numeric(argv[9]) | |
| 193 fdrThresh <- as.numeric(argv[10]) | |
| 194 lfcThresh <- as.numeric(argv[11]) | |
| 195 workMode <- as.character(argv[12]) | |
| 196 htmlPath <- as.character(argv[13]) | |
| 197 folderPath <- as.character(argv[14]) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
198 |
| 2 | 199 if (workMode=="classic") { |
| 200 pairData <- character() | |
| 201 pairData[2] <- as.character(argv[15]) | |
| 202 pairData[1] <- as.character(argv[16]) | |
| 203 } else if (workMode=="glm") { | |
| 204 contrastData <- as.character(argv[15]) | |
| 205 roastOpt <- as.character(argv[16]) | |
| 206 hairpinReq <- as.numeric(argv[17]) | |
| 207 selectOpt <- as.character(argv[18]) | |
| 208 selectVals <- as.character(argv[19]) | |
| 209 } | |
| 210 | |
| 211 # Read in inputs | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
212 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
213 samples <- read.table(samplePath, header=TRUE, sep="\t") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
214 anno <- read.table(annoPath, header=TRUE, sep="\t") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
215 if (inputType=="counts") { |
| 2 | 216 counts <- read.table(countPath, header=TRUE, sep="\t") |
| 217 } | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
218 |
| 2 | 219 ###################### Check inputs for correctness ############################ |
| 220 samples$ID <- make.names(samples$ID) | |
| 221 | |
| 222 if (!any(grepl("group", names(samples)))) { | |
| 223 stop("'group' column not specified in sample annotation file") | |
| 224 } # Check if grouping variable has been specified | |
| 225 | |
| 226 if (any(table(samples$ID)>1)){ | |
| 227 tab <- table(samples$ID) | |
| 228 offenders <- paste(names(tab[tab>1]), collapse=", ") | |
| 229 offenders <- unmake.names(offenders) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
230 stop("'ID' column of sample annotation must have unique values, values ", |
| 2 | 231 offenders, " are repeated") |
| 232 } # Check that IDs in sample annotation are unique | |
| 233 | |
| 234 if (inputType=="fastq") { | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
235 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
236 if (any(table(anno$ID)>1)){ |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
237 tab <- table(anno$ID) |
| 2 | 238 offenders <- paste(names(tab[tab>1]), collapse=", ") |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
239 stop("'ID' column of hairpin annotation must have unique values, values ", |
| 2 | 240 offenders, " are repeated") |
| 241 } # Check that IDs in hairpin annotation are unique | |
|
4
f8af57d6f60b
Fixed bug causing fastq input to break
shian_su <registertonysu@gmail.com>
parents:
2
diff
changeset
|
242 |
| 2 | 243 } else if (inputType=="counts") { |
|
4
f8af57d6f60b
Fixed bug causing fastq input to break
shian_su <registertonysu@gmail.com>
parents:
2
diff
changeset
|
244 if (any(is.na(match(samples$ID, colnames(counts))))) { |
|
f8af57d6f60b
Fixed bug causing fastq input to break
shian_su <registertonysu@gmail.com>
parents:
2
diff
changeset
|
245 stop("not all samples have groups specified") |
|
f8af57d6f60b
Fixed bug causing fastq input to break
shian_su <registertonysu@gmail.com>
parents:
2
diff
changeset
|
246 } # Check that a group has be specifed for each sample |
| 2 | 247 |
| 248 if (any(table(counts$ID)>1)){ | |
| 249 tab <- table(counts$ID) | |
| 250 offenders <- paste(names(tab[tab>1]), collapse=", ") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
251 stop("'ID' column of count table must have unique values, values ", |
| 2 | 252 offenders, " are repeated") |
| 253 } # Check that IDs in count table are unique | |
| 254 } | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
255 if (workMode=="glm") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
256 if (roastOpt == "yes") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
257 if (is.na(match("Gene", colnames(anno)))) { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
258 tempStr <- paste("Gene-wise tests selected but'Gene' column not", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
259 "specified in hairpin annotation file") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
260 stop(tempStr) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
261 } |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
262 } |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
263 } |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
264 |
| 2 | 265 ################################################################################ |
| 266 | |
| 267 # Process arguments | |
| 268 if (workMode=="glm") { | |
| 269 if (roastOpt=="yes") { | |
| 270 wantRoast <- TRUE | |
| 271 } else { | |
| 272 wantRoast <- FALSE | |
| 273 } | |
| 274 } | |
| 275 | |
| 276 # Split up contrasts seperated by comma into a vector and replace spaces with | |
| 277 # periods | |
| 278 if (exists("contrastData")) { | |
| 279 contrastData <- unlist(strsplit(contrastData, split=",")) | |
| 280 contrastData <- sanitiseEquation(contrastData) | |
| 281 contrastData <- gsub(" ", ".", contrastData, fixed=TRUE) | |
| 282 } | |
| 283 | |
| 284 # Replace spaces with periods in pair data | |
| 285 if (exists("pairData")) { | |
| 286 pairData <- make.names(pairData) | |
| 287 } | |
| 288 | |
| 289 # Generate output folder and paths | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
290 dir.create(folderPath, showWarnings=FALSE) |
| 2 | 291 |
| 292 # Generate links for outputs | |
| 293 imgOut("barHairpin") | |
| 294 imgOut("barIndex") | |
| 295 imgOut("mds") | |
| 296 imgOut("bcv") | |
| 297 if (workMode == "classic") { | |
| 298 smearPng <- makeOut(paste0("smear(", pairData[2], "-", pairData[1],").png")) | |
| 299 smearPdf <- makeOut(paste0("smear(", pairData[2], "-", pairData[1],").pdf")) | |
| 300 topOut <- makeOut(paste0("toptag(", pairData[2], "-", pairData[1],").tsv")) | |
| 301 } else if (workMode=="glm") { | |
| 302 smearPng <- character() | |
| 303 smearPdf <- character() | |
| 304 topOut <- character() | |
| 305 roastOut <- character() | |
| 306 barcodePng <- character() | |
| 307 barcodePdf <- character() | |
| 308 for (i in 1:length(contrastData)) { | |
| 309 smearPng[i] <- makeOut(paste0("smear(", contrastData[i], ").png")) | |
| 310 smearPdf[i] <- makeOut(paste0("smear(", contrastData[i], ").pdf")) | |
| 311 topOut[i] <- makeOut(paste0("toptag(", contrastData[i], ").tsv")) | |
| 8 | 312 roastOut[i] <- makeOut(paste0("gene_level(", contrastData[i], ").tsv")) |
| 2 | 313 barcodePng[i] <- makeOut(paste0("barcode(", contrastData[i], ").png")) |
| 314 barcodePdf[i] <- makeOut(paste0("barcode(", contrastData[i], ").pdf")) | |
| 315 } | |
| 316 } | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
317 countsOut <- makeOut("counts.tsv") |
| 8 | 318 sessionOut <- makeOut("session_info.txt") |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
319 |
| 2 | 320 # Initialise data for html links and images, table with the link label and |
| 321 # link address | |
| 322 linkData <- data.frame(Label=character(), Link=character(), | |
| 323 stringsAsFactors=FALSE) | |
| 324 imageData <- data.frame(Label=character(), Link=character(), | |
| 325 stringsAsFactors=FALSE) | |
| 7 | 326 |
| 327 # Initialise vectors for storage of up/down/neutral regulated counts | |
| 328 upCount <- numeric() | |
| 329 downCount <- numeric() | |
| 330 flatCount <- numeric() | |
| 331 | |
| 2 | 332 ################################################################################ |
| 333 ### Data Processing | |
| 334 ################################################################################ | |
| 335 | |
| 336 # Transform gene selection from string into index values for mroast | |
| 337 if (workMode=="glm") { | |
| 338 if (selectOpt=="rank") { | |
| 339 selectVals <- gsub(" ", "", selectVals, fixed=TRUE) | |
| 340 selectVals <- unlist(strsplit(selectVals, ",")) | |
| 341 | |
| 342 for (i in 1:length(selectVals)) { | |
| 343 if (grepl(":", selectVals[i], fixed=TRUE)) { | |
| 344 temp <- unlist(strsplit(selectVals[i], ":")) | |
| 345 selectVals <- selectVals[-i] | |
| 346 a <- as.numeric(temp[1]) | |
| 347 b <- as.numeric(temp[2]) | |
| 348 selectVals <- c(selectVals, a:b) | |
| 349 } | |
| 350 } | |
| 351 selectVals <- as.numeric(unique(selectVals)) | |
| 352 } else { | |
| 353 selectVals <- gsub(" ", "", selectVals, fixed=TRUE) | |
| 8 | 354 selectVals <- unlist(strsplit(selectVals, ",")) |
| 2 | 355 } |
| 356 } | |
| 357 | |
| 358 if (inputType=="fastq") { | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
359 # Use EdgeR hairpin process and capture outputs |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
360 hpReadout <- capture.output( |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
361 data <- processHairpinReads(fastqPath, samplePath, annoPath, |
| 2 | 362 hairpinStart=hpStart, hairpinEnd=hpEnd, |
| 363 verbose=TRUE) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
364 ) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
365 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
366 # Remove function output entries that show processing data or is empty |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
367 hpReadout <- hpReadout[hpReadout!=""] |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
368 hpReadout <- hpReadout[!grepl("Processing", hpReadout)] |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
369 hpReadout <- hpReadout[!grepl("in file", hpReadout)] |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
370 hpReadout <- gsub(" -- ", "", hpReadout, fixed=TRUE) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
371 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
372 # Make the names of groups syntactically valid (replace spaces with periods) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
373 data$samples$group <- make.names(data$samples$group) |
|
4
f8af57d6f60b
Fixed bug causing fastq input to break
shian_su <registertonysu@gmail.com>
parents:
2
diff
changeset
|
374 } else if (inputType=="counts") { |
| 2 | 375 # Process counts information, set ID column to be row names |
| 376 rownames(counts) <- counts$ID | |
| 377 counts <- counts[ , !(colnames(counts)=="ID")] | |
| 378 countsRows <- nrow(counts) | |
| 379 | |
| 380 # Process group information | |
| 381 factors <- samples$group[match(samples$ID, colnames(counts))] | |
| 382 annoRows <- nrow(anno) | |
| 383 anno <- anno[match(rownames(counts), anno$ID), ] | |
| 384 annoMatched <- sum(!is.na(anno$ID)) | |
| 385 | |
| 386 if (any(is.na(anno$ID))) { | |
| 387 warningStr <- paste("count table contained more hairpins than", | |
| 388 "specified in hairpin annotation file") | |
| 389 warning(warningStr) | |
| 390 } | |
| 391 | |
| 392 # Filter out rows with zero counts | |
| 393 sel <- rowSums(counts)!=0 | |
| 394 counts <- counts[sel, ] | |
| 395 anno <- anno[sel, ] | |
| 396 | |
| 397 # Create DGEList | |
| 398 data <- DGEList(counts=counts, lib.size=colSums(counts), | |
| 399 norm.factors=rep(1,ncol(counts)), genes=anno, group=factors) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
400 |
| 2 | 401 # Make the names of groups syntactically valid (replace spaces with periods) |
| 402 data$samples$group <- make.names(data$samples$group) | |
| 403 } | |
| 404 | |
| 405 # Filter hairpins with low counts | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
406 preFilterCount <- nrow(data) |
| 2 | 407 sel <- rowSums(cpm(data$counts) > cpmReq) >= sampleReq |
| 408 data <- data[sel, ] | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
409 postFilterCount <- nrow(data) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
410 filteredCount <- preFilterCount-postFilterCount |
| 2 | 411 |
| 412 # Estimate dispersions | |
| 413 data <- estimateDisp(data) | |
| 414 commonBCV <- sqrt(data$common.dispersion) | |
| 415 | |
| 416 ################################################################################ | |
| 417 ### Output Processing | |
| 418 ################################################################################ | |
| 419 | |
| 420 # Plot number of hairpins that could be matched per sample | |
| 421 png(barIndexPng, width=600, height=600) | |
| 422 barplot(height<-colSums(data$counts), las=2, main="Counts per index", | |
| 423 cex.names=1.0, cex.axis=0.8, ylim=c(0, max(height)*1.2)) | |
| 424 imageData[1, ] <- c("Counts per Index", "barIndex.png") | |
| 425 invisible(dev.off()) | |
| 426 | |
| 427 pdf(barIndexPdf) | |
| 428 barplot(height<-colSums(data$counts), las=2, main="Counts per index", | |
| 429 cex.names=1.0, cex.axis=0.8, ylim=c(0, max(height)*1.2)) | |
| 430 linkData[1, ] <- c("Counts per Index Barplot (.pdf)", "barIndex.pdf") | |
| 431 invisible(dev.off()) | |
| 432 | |
| 433 # Plot per hairpin totals across all samples | |
| 434 png(barHairpinPng, width=600, height=600) | |
| 435 if (nrow(data$counts)<50) { | |
| 436 barplot(height<-rowSums(data$counts), las=2, main="Counts per hairpin", | |
| 437 cex.names=0.8, cex.axis=0.8, ylim=c(0, max(height)*1.2)) | |
| 438 } else { | |
| 439 barplot(height<-rowSums(data$counts), las=2, main="Counts per hairpin", | |
| 440 cex.names=0.8, cex.axis=0.8, ylim=c(0, max(height)*1.2), | |
| 441 names.arg=FALSE) | |
| 442 } | |
| 443 imageData <- rbind(imageData, c("Counts per Hairpin", "barHairpin.png")) | |
| 444 invisible(dev.off()) | |
| 445 | |
| 446 pdf(barHairpinPdf) | |
| 447 if (nrow(data$counts)<50) { | |
| 448 barplot(height<-rowSums(data$counts), las=2, main="Counts per hairpin", | |
| 449 cex.names=0.8, cex.axis=0.8, ylim=c(0, max(height)*1.2)) | |
| 450 } else { | |
| 451 barplot(height<-rowSums(data$counts), las=2, main="Counts per hairpin", | |
| 452 cex.names=0.8, cex.axis=0.8, ylim=c(0, max(height)*1.2), | |
| 453 names.arg=FALSE) | |
| 454 } | |
| 455 newEntry <- c("Counts per Hairpin Barplot (.pdf)", "barHairpin.pdf") | |
| 456 linkData <- rbind(linkData, newEntry) | |
| 457 invisible(dev.off()) | |
| 458 | |
| 459 # Make an MDS plot to visualise relationships between replicate samples | |
| 460 png(mdsPng, width=600, height=600) | |
| 461 plotMDS(data, labels=data$samples$group, col=as.numeric(data$samples$group), | |
| 462 main="MDS Plot") | |
| 463 imageData <- rbind(imageData, c("MDS Plot", "mds.png")) | |
| 464 invisible(dev.off()) | |
| 465 | |
| 466 pdf(mdsPdf) | |
| 467 plotMDS(data, labels=data$samples$group, col=as.numeric(data$samples$group), | |
| 468 main="MDS Plot") | |
| 469 newEntry <- c("MDS Plot (.pdf)", "mds.pdf") | |
| 470 linkData <- rbind(linkData, newEntry) | |
| 471 invisible(dev.off()) | |
| 472 | |
| 473 if (workMode=="classic") { | |
| 474 # Assess differential representation using classic exact testing methodology | |
| 475 # in edgeR | |
| 476 testData <- exactTest(data, pair=pairData) | |
| 477 | |
| 478 top <- topTags(testData, n=Inf) | |
| 479 topIDs <- top$table[(top$table$FDR < fdrThresh) & | |
| 480 (abs(top$table$logFC) > lfcThresh), 1] | |
| 7 | 481 |
| 2 | 482 write.table(top, file=topOut, row.names=FALSE, sep="\t") |
| 7 | 483 |
| 2 | 484 linkName <- paste0("Top Tags Table(", pairData[2], "-", pairData[1], |
| 485 ") (.tsv)") | |
| 486 linkAddr <- paste0("toptag(", pairData[2], "-", pairData[1], ").tsv") | |
| 487 linkData <- rbind(linkData, c(linkName, linkAddr)) | |
| 488 | |
| 7 | 489 upCount[1] <- sum(top$table$FDR < fdrThresh & top$table$logFC > lfcThresh) |
| 490 downCount[1] <- sum(top$table$FDR < fdrThresh & | |
| 491 top$table$logFC < -lfcThresh) | |
| 492 flatCount[1] <- sum(top$table$FDR > fdrThresh | | |
| 493 abs(top$table$logFC) < lfcThresh) | |
| 494 | |
| 495 | |
| 496 | |
| 2 | 497 # Select hairpins with FDR < 0.05 to highlight on plot |
| 498 png(smearPng, width=600, height=600) | |
| 499 plotTitle <- gsub(".", " ", | |
| 500 paste0("Smear Plot: ", pairData[2], "-", pairData[1]), | |
| 7 | 501 fixed=TRUE) |
| 2 | 502 plotSmear(testData, de.tags=topIDs, |
| 503 pch=20, cex=1.0, main=plotTitle) | |
| 7 | 504 abline(h=c(-1, 0, 1), col=c("dodgerblue", "yellow", "dodgerblue"), lty=2) |
| 2 | 505 imgName <- paste0("Smear Plot(", pairData[2], "-", pairData[1], ")") |
| 506 imgAddr <- paste0("smear(", pairData[2], "-", pairData[1],").png") | |
| 507 imageData <- rbind(imageData, c(imgName, imgAddr)) | |
| 508 invisible(dev.off()) | |
| 509 | |
| 510 pdf(smearPdf) | |
| 511 plotTitle <- gsub(".", " ", | |
| 512 paste0("Smear Plot: ", pairData[2], "-", pairData[1]), | |
| 7 | 513 fixed=TRUE) |
| 2 | 514 plotSmear(testData, de.tags=topIDs, |
| 515 pch=20, cex=1.0, main=plotTitle) | |
| 7 | 516 abline(h=c(-1, 0, 1), col=c("dodgerblue", "yellow", "dodgerblue"), lty=2) |
| 2 | 517 imgName <- paste0("Smear Plot(", pairData[2], "-", pairData[1], ") (.pdf)") |
| 518 imgAddr <- paste0("smear(", pairData[2], "-", pairData[1], ").pdf") | |
| 519 linkData <- rbind(linkData, c(imgName, imgAddr)) | |
| 520 invisible(dev.off()) | |
| 7 | 521 |
| 2 | 522 } else if (workMode=="glm") { |
| 523 # Generating design information | |
| 524 factors <- factor(data$sample$group) | |
| 525 design <- model.matrix(~0+factors) | |
| 526 | |
| 527 colnames(design) <- gsub("factors", "", colnames(design), fixed=TRUE) | |
| 528 | |
| 529 # Split up contrasts seperated by comma into a vector | |
| 530 contrastData <- unlist(strsplit(contrastData, split=",")) | |
| 7 | 531 |
| 2 | 532 for (i in 1:length(contrastData)) { |
| 533 # Generate contrasts information | |
| 534 contrasts <- makeContrasts(contrasts=contrastData[i], levels=design) | |
| 535 | |
| 536 # Fit negative bionomial GLM | |
| 7 | 537 fit <- glmFit(data, design) |
| 2 | 538 # Carry out Likelihood ratio test |
| 7 | 539 testData <- glmLRT(fit, contrast=contrasts) |
| 2 | 540 |
| 541 # Select hairpins with FDR < 0.05 to highlight on plot | |
| 542 top <- topTags(testData, n=Inf) | |
| 543 topIDs <- top$table[(top$table$FDR < fdrThresh) & | |
| 544 (abs(top$table$logFC) > lfcThresh), 1] | |
| 545 write.table(top, file=topOut[i], row.names=FALSE, sep="\t") | |
| 546 | |
| 547 linkName <- paste0("Top Tags Table(", contrastData[i], ") (.tsv)") | |
| 548 linkAddr <- paste0("toptag(", contrastData[i], ").tsv") | |
| 549 linkData <- rbind(linkData, c(linkName, linkAddr)) | |
| 550 | |
| 7 | 551 # Collect counts for differential representation |
| 552 upCount[i] <- sum(top$table$FDR < fdrThresh & top$table$logFC > lfcThresh) | |
| 553 downCount[i] <- sum(top$table$FDR < fdrThresh & | |
| 554 top$table$logFC < -lfcThresh) | |
| 555 flatCount[i] <- sum(top$table$FDR > fdrThresh | | |
| 556 abs(top$table$logFC) < lfcThresh) | |
| 557 | |
| 2 | 558 # Make a plot of logFC versus logCPM |
| 559 png(smearPng[i], height=600, width=600) | |
| 560 plotTitle <- paste("Smear Plot:", gsub(".", " ", contrastData[i], | |
| 561 fixed=TRUE)) | |
| 562 plotSmear(testData, de.tags=topIDs, pch=20, cex=0.8, main=plotTitle) | |
| 563 abline(h=c(-1, 0, 1), col=c("dodgerblue", "yellow", "dodgerblue"), lty=2) | |
| 564 | |
| 565 imgName <- paste0("Smear Plot(", contrastData[i], ")") | |
| 566 imgAddr <- paste0("smear(", contrastData[i], ").png") | |
| 567 imageData <- rbind(imageData, c(imgName, imgAddr)) | |
| 568 invisible(dev.off()) | |
| 569 | |
| 570 pdf(smearPdf[i]) | |
| 571 plotTitle <- paste("Smear Plot:", gsub(".", " ", contrastData[i], | |
| 572 fixed=TRUE)) | |
| 573 plotSmear(testData, de.tags=topIDs, pch=20, cex=0.8, main=plotTitle) | |
| 574 abline(h=c(-1, 0, 1), col=c("dodgerblue", "yellow", "dodgerblue"), lty=2) | |
| 575 | |
| 576 linkName <- paste0("Smear Plot(", contrastData[i], ") (.pdf)") | |
| 577 linkAddr <- paste0("smear(", contrastData[i], ").pdf") | |
| 578 linkData <- rbind(linkData, c(linkName, linkAddr)) | |
| 579 invisible(dev.off()) | |
| 580 | |
| 581 genes <- as.character(data$genes$Gene) | |
| 582 unq <- unique(genes) | |
| 583 unq <- unq[!is.na(unq)] | |
| 584 geneList <- list() | |
| 585 for (gene in unq) { | |
| 586 if (length(which(genes==gene)) >= hairpinReq) { | |
| 587 geneList[[gene]] <- which(genes==gene) | |
| 588 } | |
| 589 } | |
| 590 | |
| 591 if (wantRoast) { | |
| 592 # Input preparaton for roast | |
| 7 | 593 nrot <- 9999 |
| 2 | 594 set.seed(602214129) |
| 595 roastData <- mroast(data, index=geneList, design=design, | |
| 596 contrast=contrasts, nrot=nrot) | |
| 597 roastData <- cbind(GeneID=rownames(roastData), roastData) | |
| 598 write.table(roastData, file=roastOut[i], row.names=FALSE, sep="\t") | |
| 599 linkName <- paste0("Gene Level Analysis Table(", contrastData[i], | |
| 600 ") (.tsv)") | |
| 8 | 601 linkAddr <- paste0("gene_level(", contrastData[i], ").tsv") |
| 2 | 602 linkData <- rbind(linkData, c(linkName, linkAddr)) |
| 603 if (selectOpt=="rank") { | |
| 604 selectedGenes <- rownames(roastData)[selectVals] | |
| 605 } else { | |
| 606 selectedGenes <- selectVals | |
| 607 } | |
| 608 | |
| 609 if (packageVersion("limma")<"3.19.19") { | |
| 610 png(barcodePng[i], width=600, height=length(selectedGenes)*150) | |
| 611 } else { | |
| 612 png(barcodePng[i], width=600, height=length(selectedGenes)*300) | |
| 613 } | |
| 614 par(mfrow=c(length(selectedGenes), 1)) | |
| 615 for (gene in selectedGenes) { | |
| 616 barcodeplot(testData$table$logFC, index=geneList[[gene]], | |
| 617 main=paste("Barcode Plot for", gene, "(logFCs)", | |
| 618 gsub(".", " ", contrastData[i])), | |
| 619 labels=c("Positive logFC", "Negative logFC")) | |
| 620 } | |
| 621 imgName <- paste0("Barcode Plot(", contrastData[i], ")") | |
| 622 imgAddr <- paste0("barcode(", contrastData[i], ").png") | |
| 623 imageData <- rbind(imageData, c(imgName, imgAddr)) | |
| 624 dev.off() | |
| 625 if (packageVersion("limma")<"3.19.19") { | |
| 626 pdf(barcodePdf[i], width=8, height=2) | |
| 627 } else { | |
| 628 pdf(barcodePdf[i], width=8, height=4) | |
| 629 } | |
| 630 for (gene in selectedGenes) { | |
| 631 barcodeplot(testData$table$logFC, index=geneList[[gene]], | |
| 632 main=paste("Barcode Plot for", gene, "(logFCs)", | |
| 633 gsub(".", " ", contrastData[i])), | |
| 634 labels=c("Positive logFC", "Negative logFC")) | |
| 635 } | |
| 636 linkName <- paste0("Barcode Plot(", contrastData[i], ") (.pdf)") | |
| 637 linkAddr <- paste0("barcode(", contrastData[i], ").pdf") | |
| 638 linkData <- rbind(linkData, c(linkName, linkAddr)) | |
| 639 dev.off() | |
| 640 } | |
| 641 } | |
| 642 } | |
| 643 | |
| 8 | 644 # Generate data frame of the significant differences |
| 7 | 645 sigDiff <- data.frame(Up=upCount, Flat=flatCount, Down=downCount) |
| 646 if (workMode == "glm") { | |
| 647 row.names(sigDiff) <- contrastData | |
| 648 } else if (workMode == "classic") { | |
| 649 row.names(sigDiff) <- paste0(pairData[2], "-", pairData[1]) | |
| 650 } | |
| 651 | |
| 8 | 652 # Output table of summarised counts |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
653 ID <- rownames(data$counts) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
654 outputCounts <- cbind(ID, data$counts) |
| 8 | 655 write.table(outputCounts, file=countsOut, row.names=FALSE, sep="\t", |
| 656 quote=FALSE) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
657 linkName <- "Counts table (.tsv)" |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
658 linkAddr <- "counts.tsv" |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
659 linkData <- rbind(linkData, c(linkName, linkAddr)) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
660 |
| 8 | 661 # Record session info |
| 662 writeLines(capture.output(sessionInfo()), sessionOut) | |
| 663 linkData <- rbind(linkData, c("Session Info", "session_info.txt")) | |
| 664 | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
665 # Record ending time and calculate total run time |
| 2 | 666 timeEnd <- as.character(Sys.time()) |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
667 timeTaken <- capture.output(round(difftime(timeEnd,timeStart), digits=3)) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
668 timeTaken <- gsub("Time difference of ", "", timeTaken, fixed=TRUE) |
| 2 | 669 ################################################################################ |
| 670 ### HTML Generation | |
| 671 ################################################################################ | |
| 672 # Clear file | |
| 673 cat("", file=htmlPath) | |
| 674 | |
| 675 cata("<html>\n") | |
| 676 HtmlHead("EdgeR Output") | |
| 677 | |
| 678 cata("<body>\n") | |
| 679 cata("<h3>EdgeR Analysis Output:</h3>\n") | |
| 680 cata("<h4>Input Summary:</h4>\n") | |
| 681 if (inputType=="fastq") { | |
| 682 cata("<ul>\n") | |
| 683 ListItem(hpReadout[1]) | |
| 684 ListItem(hpReadout[2]) | |
| 685 cata("</ul>\n") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
686 cata(hpReadout[3], "<br />\n") |
| 2 | 687 cata("<ul>\n") |
| 688 ListItem(hpReadout[4]) | |
| 689 ListItem(hpReadout[7]) | |
| 690 cata("</ul>\n") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
691 cata(hpReadout[8:11], sep="<br />\n") |
| 2 | 692 cata("<br />\n") |
| 693 cata("<b>Please check that read percentages are consistent with ") | |
| 694 cata("expectations.</b><br >\n") | |
| 695 } else if (inputType=="counts") { | |
| 696 cata("<ul>\n") | |
| 697 ListItem("Number of Samples: ", ncol(data$counts)) | |
| 698 ListItem("Number of Hairpins: ", countsRows) | |
| 699 ListItem("Number of annotations provided: ", annoRows) | |
| 700 ListItem("Number of annotations matched to hairpin: ", annoMatched) | |
| 701 cata("</ul>\n") | |
| 702 } | |
| 703 | |
| 704 cata("The estimated common biological coefficient of variation (BCV) is: ", | |
| 705 commonBCV, "<br />\n") | |
| 706 | |
| 707 cata("<h4>Output:</h4>\n") | |
| 7 | 708 cata("PDF copies of JPEGS available in 'Plots' section.<br />\n") |
| 2 | 709 for (i in 1:nrow(imageData)) { |
| 710 if (grepl("barcode", imageData$Link[i])) { | |
| 711 if (packageVersion("limma")<"3.19.19") { | |
| 712 HtmlImage(imageData$Link[i], imageData$Label[i], | |
| 713 height=length(selectedGenes)*150) | |
| 714 } else { | |
| 715 HtmlImage(imageData$Link[i], imageData$Label[i], | |
| 716 height=length(selectedGenes)*300) | |
| 717 } | |
| 718 } else { | |
| 719 HtmlImage(imageData$Link[i], imageData$Label[i]) | |
| 720 } | |
| 721 } | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
722 cata("<br />\n") |
| 2 | 723 |
| 7 | 724 cata("<h4>Differential Representation Counts:</h4>\n") |
| 725 | |
| 726 cata("<table border=\"1\" cellpadding=\"4\">\n") | |
| 727 cata("<tr>\n") | |
| 728 TableItem() | |
| 729 for (i in colnames(sigDiff)) { | |
| 730 TableHeadItem(i) | |
| 731 } | |
| 732 cata("</tr>\n") | |
| 733 for (i in 1:nrow(sigDiff)) { | |
| 734 cata("<tr>\n") | |
| 735 TableHeadItem(unmake.names(row.names(sigDiff)[i])) | |
| 736 for (j in 1:ncol(sigDiff)) { | |
| 737 TableItem(as.character(sigDiff[i, j])) | |
| 738 } | |
| 739 cata("</tr>\n") | |
| 740 } | |
| 741 cata("</table>") | |
| 742 | |
| 2 | 743 cata("<h4>Plots:</h4>\n") |
| 744 for (i in 1:nrow(linkData)) { | |
| 8 | 745 if (grepl(".pdf", linkData$Link[i])) { |
| 2 | 746 HtmlLink(linkData$Link[i], linkData$Label[i]) |
| 747 } | |
| 748 } | |
| 749 | |
| 750 cata("<h4>Tables:</h4>\n") | |
| 751 for (i in 1:nrow(linkData)) { | |
| 752 if (grepl(".tsv", linkData$Link[i])) { | |
| 753 HtmlLink(linkData$Link[i], linkData$Label[i]) | |
| 754 } | |
| 755 } | |
| 756 | |
| 7 | 757 cata("<p>Alt-click links to download file.</p>\n") |
| 758 cata("<p>Click floppy disc icon associated history item to download ") | |
| 759 cata("all files.</p>\n") | |
| 760 cata("<p>.tsv files can be viewed in Excel or any spreadsheet program.</p>\n") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
761 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
762 cata("<h4>Additional Information:</h4>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
763 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
764 if (inputType == "fastq") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
765 ListItem("Data was gathered from fastq raw read file(s).") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
766 } else if (inputType == "counts") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
767 ListItem("Data was gathered from a table of counts.") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
768 } |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
769 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
770 if (cpmReq!=0 && sampleReq!=0) { |
| 8 | 771 tempStr <- paste("Hairpins without more than", cpmReq, |
| 7 | 772 "CPM in at least", sampleReq, "samples are insignificant", |
| 773 "and filtered out.") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
774 ListItem(tempStr) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
775 filterProp <- round(filteredCount/preFilterCount*100, digits=2) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
776 tempStr <- paste0(filteredCount, " of ", preFilterCount," (", filterProp, |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
777 "%) hairpins were filtered out for low count-per-million.") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
778 ListItem(tempStr) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
779 } |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
780 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
781 if (workMode == "classic") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
782 ListItem("An exact test was performed on each hairpin.") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
783 } else if (workMode == "glm") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
784 ListItem("A generalised linear model was fitted to each hairpin.") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
785 } |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
786 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
787 cit <- character() |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
788 link <-character() |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
789 link[1] <- paste0("<a href=\"", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
790 "http://www.bioconductor.org/packages/release/bioc/", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
791 "vignettes/limma/inst/doc/usersguide.pdf", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
792 "\">", "limma User's Guide", "</a>.") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
793 link[2] <- paste0("<a href=\"", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
794 "http://www.bioconductor.org/packages/release/bioc/", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
795 "vignettes/edgeR/inst/doc/edgeRUsersGuide.pdf", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
796 "\">", "edgeR User's Guide", "</a>") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
797 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
798 cit[1] <- paste("Robinson MD, McCarthy DJ and Smyth GK (2010).", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
799 "edgeR: a Bioconductor package for differential", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
800 "expression analysis of digital gene expression", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
801 "data. Bioinformatics 26, 139-140") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
802 cit[2] <- paste("Robinson MD and Smyth GK (2007). Moderated statistical tests", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
803 "for assessing differences in tag abundance. Bioinformatics", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
804 "23, 2881-2887") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
805 cit[3] <- paste("Robinson MD and Smyth GK (2008). Small-sample estimation of", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
806 "negative binomial dispersion, with applications to SAGE data.", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
807 "Biostatistics, 9, 321-332") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
808 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
809 cit[4] <- paste("McCarthy DJ, Chen Y and Smyth GK (2012). Differential", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
810 "expression analysis of multifactor RNA-Seq experiments with", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
811 "respect to biological variation. Nucleic Acids Research 40,", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
812 "4288-4297") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
813 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
814 cata("<h4>Citations</h4>") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
815 cata("<ol>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
816 ListItem(cit[1]) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
817 ListItem(cit[2]) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
818 ListItem(cit[3]) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
819 ListItem(cit[4]) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
820 cata("</ol>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
821 |
| 8 | 822 cata("<p>Report problems to: su.s@wehi.edu.au</p>\n") |
| 823 | |
| 824 for (i in 1:nrow(linkData)) { | |
| 825 if (grepl("session_info", linkData$Link[i])) { | |
| 826 HtmlLink(linkData$Link[i], linkData$Label[i]) | |
| 827 } | |
| 828 } | |
| 829 | |
| 2 | 830 cata("<table border=\"0\">\n") |
| 831 cata("<tr>\n") | |
| 832 TableItem("Task started at:"); TableItem(timeStart) | |
| 833 cata("</tr>\n") | |
| 834 cata("<tr>\n") | |
| 835 TableItem("Task ended at:"); TableItem(timeEnd) | |
| 836 cata("</tr>\n") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
837 cata("<tr>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
838 TableItem("Task run time:"); TableItem(timeTaken) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
839 cata("<tr>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
840 cata("</table>\n") |
| 2 | 841 |
| 842 cata("</body>\n") | |
| 843 cata("</html>") |