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