Mercurial > repos > shians > shrnaseq
annotate hairpinTool.R @ 10:8923d4ea858b
- Added check for zero library size, will now filter out zero library size
samples and generate report of filtered samples in html output
| author | shian_su <registertonysu@gmail.com> |
|---|---|
| date | Tue, 12 Aug 2014 14:42:27 +1000 |
| parents | f1076bfb0ed1 |
| children | c0a76e30d61b |
| 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 |
|
9
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
38 # ### |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
39 # |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
40 # OUT: Bar Plot of Counts Per Index |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
41 # Bar Plot of Counts Per Hairpin |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
42 # MDS Plot |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
43 # Smear Plot |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
44 # Barcode Plots (If Genewise testing was selected) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
45 # Top Expression Table |
| 7 | 46 # Feature Counts Table |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
47 # HTML file linking to the ouputs |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
48 # |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
49 # Author: Shian Su - registertonysu@gmail.com - Jan 2014 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
50 |
| 2 | 51 # Record starting time |
| 52 timeStart <- as.character(Sys.time()) | |
| 53 | |
| 54 # Loading and checking required packages | |
| 55 library(methods, quietly=TRUE, warn.conflicts=FALSE) | |
| 56 library(statmod, quietly=TRUE, warn.conflicts=FALSE) | |
| 57 library(splines, quietly=TRUE, warn.conflicts=FALSE) | |
| 58 library(edgeR, quietly=TRUE, warn.conflicts=FALSE) | |
| 59 library(limma, quietly=TRUE, warn.conflicts=FALSE) | |
| 60 | |
|
9
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
61 if (packageVersion("edgeR") < "3.5.27") { |
| 8 | 62 stop("Please update 'edgeR' to version >= 3.5.23 to run this tool") |
| 63 } | |
| 64 | |
| 65 if (packageVersion("limma")<"3.19.19") { | |
| 66 message("Update 'limma' to version >= 3.19.19 to see updated barcode graphs") | |
| 2 | 67 } |
| 68 | |
| 69 ################################################################################ | |
| 70 ### Function declarations | |
| 71 ################################################################################ | |
| 72 | |
| 7 | 73 # Function to load libaries without messages |
| 74 silentLibrary <- function(...) { | |
| 75 list <- c(...) | |
| 76 for (package in list){ | |
| 77 suppressPackageStartupMessages(library(package, character.only=TRUE)) | |
| 78 } | |
| 79 } | |
| 80 | |
| 2 | 81 # Function to sanitise contrast equations so there are no whitespaces |
| 82 # surrounding the arithmetic operators, leading or trailing whitespace | |
| 83 sanitiseEquation <- function(equation) { | |
| 84 equation <- gsub(" *[+] *", "+", equation) | |
| 85 equation <- gsub(" *[-] *", "-", equation) | |
| 86 equation <- gsub(" *[/] *", "/", equation) | |
| 87 equation <- gsub(" *[*] *", "*", equation) | |
| 88 equation <- gsub("^\\s+|\\s+$", "", equation) | |
| 89 return(equation) | |
| 90 } | |
| 91 | |
| 92 # Function to sanitise group information | |
| 93 sanitiseGroups <- function(string) { | |
| 94 string <- gsub(" *[,] *", ",", string) | |
| 95 string <- gsub("^\\s+|\\s+$", "", string) | |
| 96 return(string) | |
| 97 } | |
| 98 | |
| 99 # Function to change periods to whitespace in a string | |
| 100 unmake.names <- function(string) { | |
| 101 string <- gsub(".", " ", string, fixed=TRUE) | |
| 102 return(string) | |
| 103 } | |
| 104 | |
| 105 # Function has string input and generates an output path string | |
| 106 makeOut <- function(filename) { | |
| 107 return(paste0(folderPath, "/", filename)) | |
| 108 } | |
| 109 | |
| 110 # Function has string input and generates both a pdf and png output strings | |
| 111 imgOut <- function(filename) { | |
| 112 assign(paste0(filename, "Png"), makeOut(paste0(filename,".png")), | |
| 7 | 113 envir=.GlobalEnv) |
| 2 | 114 assign(paste0(filename, "Pdf"), makeOut(paste0(filename,".pdf")), |
| 7 | 115 envir=.GlobalEnv) |
| 2 | 116 } |
| 117 | |
| 118 # Create cat function default path set, default seperator empty and appending | |
| 119 # true by default (Ripped straight from the cat function with altered argument | |
| 120 # defaults) | |
| 7 | 121 cata <- function(..., file=htmlPath, sep="", fill=FALSE, labels=NULL, |
| 122 append=TRUE) { | |
| 2 | 123 if (is.character(file)) |
| 124 if (file == "") | |
| 125 file <- stdout() | |
| 126 else if (substring(file, 1L, 1L) == "|") { | |
| 127 file <- pipe(substring(file, 2L), "w") | |
| 128 on.exit(close(file)) | |
| 129 } | |
| 130 else { | |
| 131 file <- file(file, ifelse(append, "a", "w")) | |
| 132 on.exit(close(file)) | |
| 133 } | |
| 134 .Internal(cat(list(...), file, sep, fill, labels, append)) | |
| 135 } | |
| 136 | |
| 137 # Function to write code for html head and title | |
| 138 HtmlHead <- function(title) { | |
| 139 cata("<head>\n") | |
| 140 cata("<title>", title, "</title>\n") | |
| 141 cata("</head>\n") | |
| 142 } | |
| 143 | |
| 144 # Function to write code for html links | |
| 145 HtmlLink <- function(address, label=address) { | |
| 146 cata("<a href=\"", address, "\" target=\"_blank\">", label, "</a><br />\n") | |
| 147 } | |
| 148 | |
| 149 # Function to write code for html images | |
| 150 HtmlImage <- function(source, label=source, height=600, width=600) { | |
| 151 cata("<img src=\"", source, "\" alt=\"", label, "\" height=\"", height) | |
| 152 cata("\" width=\"", width, "\"/>\n") | |
| 153 } | |
| 154 | |
| 155 # Function to write code for html list items | |
| 156 ListItem <- function(...) { | |
| 157 cata("<li>", ..., "</li>\n") | |
| 158 } | |
| 159 | |
| 160 TableItem <- function(...) { | |
| 161 cata("<td>", ..., "</td>\n") | |
| 162 } | |
| 163 | |
| 164 TableHeadItem <- function(...) { | |
| 165 cata("<th>", ..., "</th>\n") | |
| 166 } | |
| 167 ################################################################################ | |
| 168 ### Input Processing | |
| 169 ################################################################################ | |
| 170 | |
| 171 # Grabbing arguments from command line | |
| 172 argv <- commandArgs(TRUE) | |
| 173 | |
| 174 # Remove fastq file paths after collecting from argument vector | |
| 175 inputType <- as.character(argv[1]) | |
| 176 if (inputType=="fastq") { | |
| 177 fastqPath <- as.character(gsub("fastq::", "", argv[grepl("fastq::", argv)], | |
| 178 fixed=TRUE)) | |
| 179 argv <- argv[!grepl("fastq::", argv, fixed=TRUE)] | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
180 annoPath <- as.character(argv[2]) |
| 2 | 181 samplePath <- as.character(argv[3]) |
| 182 barStart <- as.numeric(argv[4]) | |
| 183 barEnd <- as.numeric(argv[5]) | |
| 184 hpStart <- as.numeric(argv[6]) | |
| 185 hpEnd <- as.numeric(argv[7]) | |
| 186 } else if (inputType=="counts") { | |
| 187 countPath <- as.character(argv[2]) | |
| 188 annoPath <- as.character(argv[3]) | |
| 189 samplePath <- as.character(argv[4]) | |
| 190 } | |
| 191 | |
| 192 cpmReq <- as.numeric(argv[8]) | |
| 193 sampleReq <- as.numeric(argv[9]) | |
| 194 fdrThresh <- as.numeric(argv[10]) | |
| 195 lfcThresh <- as.numeric(argv[11]) | |
| 196 workMode <- as.character(argv[12]) | |
| 197 htmlPath <- as.character(argv[13]) | |
| 198 folderPath <- as.character(argv[14]) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
199 |
| 2 | 200 if (workMode=="classic") { |
| 201 pairData <- character() | |
| 202 pairData[2] <- as.character(argv[15]) | |
| 203 pairData[1] <- as.character(argv[16]) | |
| 204 } else if (workMode=="glm") { | |
| 205 contrastData <- as.character(argv[15]) | |
| 206 roastOpt <- as.character(argv[16]) | |
| 207 hairpinReq <- as.numeric(argv[17]) | |
| 208 selectOpt <- as.character(argv[18]) | |
| 209 selectVals <- as.character(argv[19]) | |
| 210 } | |
| 211 | |
| 212 # Read in inputs | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
213 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
214 samples <- read.table(samplePath, header=TRUE, sep="\t") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
215 anno <- read.table(annoPath, header=TRUE, sep="\t") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
216 if (inputType=="counts") { |
| 2 | 217 counts <- read.table(countPath, header=TRUE, sep="\t") |
| 218 } | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
219 |
| 2 | 220 ###################### Check inputs for correctness ############################ |
| 221 samples$ID <- make.names(samples$ID) | |
| 222 | |
| 223 if (!any(grepl("group", names(samples)))) { | |
| 224 stop("'group' column not specified in sample annotation file") | |
| 225 } # Check if grouping variable has been specified | |
| 226 | |
| 227 if (any(table(samples$ID)>1)){ | |
| 228 tab <- table(samples$ID) | |
| 229 offenders <- paste(names(tab[tab>1]), collapse=", ") | |
| 230 offenders <- unmake.names(offenders) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
231 stop("'ID' column of sample annotation must have unique values, values ", |
| 2 | 232 offenders, " are repeated") |
| 233 } # Check that IDs in sample annotation are unique | |
| 234 | |
| 235 if (inputType=="fastq") { | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
236 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
237 if (any(table(anno$ID)>1)){ |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
238 tab <- table(anno$ID) |
| 2 | 239 offenders <- paste(names(tab[tab>1]), collapse=", ") |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
240 stop("'ID' column of hairpin annotation must have unique values, values ", |
| 2 | 241 offenders, " are repeated") |
| 242 } # 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
|
243 |
| 2 | 244 } else if (inputType=="counts") { |
|
4
f8af57d6f60b
Fixed bug causing fastq input to break
shian_su <registertonysu@gmail.com>
parents:
2
diff
changeset
|
245 if (any(is.na(match(samples$ID, colnames(counts))))) { |
|
f8af57d6f60b
Fixed bug causing fastq input to break
shian_su <registertonysu@gmail.com>
parents:
2
diff
changeset
|
246 stop("not all samples have groups specified") |
|
f8af57d6f60b
Fixed bug causing fastq input to break
shian_su <registertonysu@gmail.com>
parents:
2
diff
changeset
|
247 } # Check that a group has be specifed for each sample |
| 2 | 248 |
| 249 if (any(table(counts$ID)>1)){ | |
| 250 tab <- table(counts$ID) | |
| 251 offenders <- paste(names(tab[tab>1]), collapse=", ") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
252 stop("'ID' column of count table must have unique values, values ", |
| 2 | 253 offenders, " are repeated") |
| 254 } # Check that IDs in count table are unique | |
| 255 } | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
256 if (workMode=="glm") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
257 if (roastOpt == "yes") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
258 if (is.na(match("Gene", colnames(anno)))) { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
259 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
|
260 "specified in hairpin annotation file") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
261 stop(tempStr) |
|
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 } |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
265 |
| 2 | 266 ################################################################################ |
| 267 | |
| 268 # Process arguments | |
| 269 if (workMode=="glm") { | |
| 270 if (roastOpt=="yes") { | |
| 271 wantRoast <- TRUE | |
| 272 } else { | |
| 273 wantRoast <- FALSE | |
| 274 } | |
| 275 } | |
| 276 | |
| 277 # Split up contrasts seperated by comma into a vector and replace spaces with | |
| 278 # periods | |
| 279 if (exists("contrastData")) { | |
| 280 contrastData <- unlist(strsplit(contrastData, split=",")) | |
| 281 contrastData <- sanitiseEquation(contrastData) | |
| 282 contrastData <- gsub(" ", ".", contrastData, fixed=TRUE) | |
| 283 } | |
| 284 | |
| 285 # Replace spaces with periods in pair data | |
| 286 if (exists("pairData")) { | |
| 287 pairData <- make.names(pairData) | |
| 288 } | |
| 289 | |
| 290 # Generate output folder and paths | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
291 dir.create(folderPath, showWarnings=FALSE) |
| 2 | 292 |
| 293 # Generate links for outputs | |
| 294 imgOut("barHairpin") | |
| 295 imgOut("barIndex") | |
| 296 imgOut("mds") | |
| 297 imgOut("bcv") | |
| 298 if (workMode == "classic") { | |
| 299 smearPng <- makeOut(paste0("smear(", pairData[2], "-", pairData[1],").png")) | |
| 300 smearPdf <- makeOut(paste0("smear(", pairData[2], "-", pairData[1],").pdf")) | |
| 301 topOut <- makeOut(paste0("toptag(", pairData[2], "-", pairData[1],").tsv")) | |
| 302 } else if (workMode=="glm") { | |
| 303 smearPng <- character() | |
| 304 smearPdf <- character() | |
| 305 topOut <- character() | |
| 306 roastOut <- character() | |
| 307 barcodePng <- character() | |
| 308 barcodePdf <- character() | |
| 309 for (i in 1:length(contrastData)) { | |
| 310 smearPng[i] <- makeOut(paste0("smear(", contrastData[i], ").png")) | |
| 311 smearPdf[i] <- makeOut(paste0("smear(", contrastData[i], ").pdf")) | |
| 312 topOut[i] <- makeOut(paste0("toptag(", contrastData[i], ").tsv")) | |
| 8 | 313 roastOut[i] <- makeOut(paste0("gene_level(", contrastData[i], ").tsv")) |
| 2 | 314 barcodePng[i] <- makeOut(paste0("barcode(", contrastData[i], ").png")) |
| 315 barcodePdf[i] <- makeOut(paste0("barcode(", contrastData[i], ").pdf")) | |
| 316 } | |
| 317 } | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
318 countsOut <- makeOut("counts.tsv") |
| 8 | 319 sessionOut <- makeOut("session_info.txt") |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
320 |
| 2 | 321 # Initialise data for html links and images, table with the link label and |
| 322 # link address | |
| 323 linkData <- data.frame(Label=character(), Link=character(), | |
| 324 stringsAsFactors=FALSE) | |
| 325 imageData <- data.frame(Label=character(), Link=character(), | |
| 326 stringsAsFactors=FALSE) | |
| 7 | 327 |
| 328 # Initialise vectors for storage of up/down/neutral regulated counts | |
| 329 upCount <- numeric() | |
| 330 downCount <- numeric() | |
| 331 flatCount <- numeric() | |
| 332 | |
| 2 | 333 ################################################################################ |
| 334 ### Data Processing | |
| 335 ################################################################################ | |
| 336 | |
| 337 # Transform gene selection from string into index values for mroast | |
| 338 if (workMode=="glm") { | |
| 339 if (selectOpt=="rank") { | |
| 340 selectVals <- gsub(" ", "", selectVals, fixed=TRUE) | |
| 341 selectVals <- unlist(strsplit(selectVals, ",")) | |
| 342 | |
| 343 for (i in 1:length(selectVals)) { | |
| 344 if (grepl(":", selectVals[i], fixed=TRUE)) { | |
| 345 temp <- unlist(strsplit(selectVals[i], ":")) | |
| 346 selectVals <- selectVals[-i] | |
| 347 a <- as.numeric(temp[1]) | |
| 348 b <- as.numeric(temp[2]) | |
| 349 selectVals <- c(selectVals, a:b) | |
| 350 } | |
| 351 } | |
| 352 selectVals <- as.numeric(unique(selectVals)) | |
| 353 } else { | |
| 354 selectVals <- gsub(" ", "", selectVals, fixed=TRUE) | |
| 8 | 355 selectVals <- unlist(strsplit(selectVals, ",")) |
| 2 | 356 } |
| 357 } | |
| 358 | |
| 359 if (inputType=="fastq") { | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
360 # Use EdgeR hairpin process and capture outputs |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
361 hpReadout <- capture.output( |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
362 data <- processHairpinReads(fastqPath, samplePath, annoPath, |
|
9
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
363 barcodeStart=barStart, barcodeEnd=barEnd, |
| 2 | 364 hairpinStart=hpStart, hairpinEnd=hpEnd, |
| 365 verbose=TRUE) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
366 ) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
367 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
368 # 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
|
369 hpReadout <- hpReadout[hpReadout!=""] |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
370 hpReadout <- hpReadout[!grepl("Processing", hpReadout)] |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
371 hpReadout <- hpReadout[!grepl("in file", hpReadout)] |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
372 hpReadout <- gsub(" -- ", "", hpReadout, fixed=TRUE) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
373 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
374 # 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
|
375 data$samples$group <- make.names(data$samples$group) |
|
4
f8af57d6f60b
Fixed bug causing fastq input to break
shian_su <registertonysu@gmail.com>
parents:
2
diff
changeset
|
376 } else if (inputType=="counts") { |
| 2 | 377 # Process counts information, set ID column to be row names |
| 378 rownames(counts) <- counts$ID | |
| 379 counts <- counts[ , !(colnames(counts)=="ID")] | |
| 380 countsRows <- nrow(counts) | |
| 381 | |
| 382 # Process group information | |
| 383 factors <- samples$group[match(samples$ID, colnames(counts))] | |
| 384 annoRows <- nrow(anno) | |
| 385 anno <- anno[match(rownames(counts), anno$ID), ] | |
| 386 annoMatched <- sum(!is.na(anno$ID)) | |
| 387 | |
| 388 if (any(is.na(anno$ID))) { | |
| 389 warningStr <- paste("count table contained more hairpins than", | |
| 390 "specified in hairpin annotation file") | |
| 391 warning(warningStr) | |
| 392 } | |
| 393 | |
| 394 # Filter out rows with zero counts | |
| 395 sel <- rowSums(counts)!=0 | |
| 396 counts <- counts[sel, ] | |
| 397 anno <- anno[sel, ] | |
| 398 | |
| 399 # Create DGEList | |
| 400 data <- DGEList(counts=counts, lib.size=colSums(counts), | |
| 401 norm.factors=rep(1,ncol(counts)), genes=anno, group=factors) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
402 |
| 2 | 403 # Make the names of groups syntactically valid (replace spaces with periods) |
| 404 data$samples$group <- make.names(data$samples$group) | |
| 405 } | |
| 406 | |
|
10
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
407 # Filter out any samples with zero counts |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
408 if (any(data$samples$lib.size == 0)) { |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
409 sampleSel <- data$samples$lib.size != 0 |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
410 filteredSamples <- paste(data$samples$ID[!sampleSel], collapse=", ") |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
411 data$counts <- data$counts[, sampleSel] |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
412 data$samples <- data$samples[sampleSel, ] |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
413 } |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
414 |
| 2 | 415 # Filter hairpins with low counts |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
416 preFilterCount <- nrow(data) |
| 2 | 417 sel <- rowSums(cpm(data$counts) > cpmReq) >= sampleReq |
| 418 data <- data[sel, ] | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
419 postFilterCount <- nrow(data) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
420 filteredCount <- preFilterCount-postFilterCount |
| 2 | 421 |
| 422 # Estimate dispersions | |
| 423 data <- estimateDisp(data) | |
| 424 commonBCV <- sqrt(data$common.dispersion) | |
| 425 | |
| 426 ################################################################################ | |
| 427 ### Output Processing | |
| 428 ################################################################################ | |
| 429 | |
| 430 # Plot number of hairpins that could be matched per sample | |
| 431 png(barIndexPng, width=600, height=600) | |
| 432 barplot(height<-colSums(data$counts), las=2, main="Counts per index", | |
| 433 cex.names=1.0, cex.axis=0.8, ylim=c(0, max(height)*1.2)) | |
| 434 imageData[1, ] <- c("Counts per Index", "barIndex.png") | |
| 435 invisible(dev.off()) | |
| 436 | |
| 437 pdf(barIndexPdf) | |
| 438 barplot(height<-colSums(data$counts), las=2, main="Counts per index", | |
| 439 cex.names=1.0, cex.axis=0.8, ylim=c(0, max(height)*1.2)) | |
| 440 linkData[1, ] <- c("Counts per Index Barplot (.pdf)", "barIndex.pdf") | |
| 441 invisible(dev.off()) | |
| 442 | |
| 443 # Plot per hairpin totals across all samples | |
| 444 png(barHairpinPng, width=600, height=600) | |
| 445 if (nrow(data$counts)<50) { | |
| 446 barplot(height<-rowSums(data$counts), las=2, main="Counts per hairpin", | |
| 447 cex.names=0.8, cex.axis=0.8, ylim=c(0, max(height)*1.2)) | |
| 448 } else { | |
| 449 barplot(height<-rowSums(data$counts), las=2, main="Counts per hairpin", | |
| 450 cex.names=0.8, cex.axis=0.8, ylim=c(0, max(height)*1.2), | |
| 451 names.arg=FALSE) | |
| 452 } | |
| 453 imageData <- rbind(imageData, c("Counts per Hairpin", "barHairpin.png")) | |
| 454 invisible(dev.off()) | |
| 455 | |
| 456 pdf(barHairpinPdf) | |
| 457 if (nrow(data$counts)<50) { | |
| 458 barplot(height<-rowSums(data$counts), las=2, main="Counts per hairpin", | |
| 459 cex.names=0.8, cex.axis=0.8, ylim=c(0, max(height)*1.2)) | |
| 460 } else { | |
| 461 barplot(height<-rowSums(data$counts), las=2, main="Counts per hairpin", | |
| 462 cex.names=0.8, cex.axis=0.8, ylim=c(0, max(height)*1.2), | |
| 463 names.arg=FALSE) | |
| 464 } | |
| 465 newEntry <- c("Counts per Hairpin Barplot (.pdf)", "barHairpin.pdf") | |
| 466 linkData <- rbind(linkData, newEntry) | |
| 467 invisible(dev.off()) | |
| 468 | |
| 469 # Make an MDS plot to visualise relationships between replicate samples | |
| 470 png(mdsPng, width=600, height=600) | |
| 471 plotMDS(data, labels=data$samples$group, col=as.numeric(data$samples$group), | |
| 472 main="MDS Plot") | |
| 473 imageData <- rbind(imageData, c("MDS Plot", "mds.png")) | |
| 474 invisible(dev.off()) | |
| 475 | |
| 476 pdf(mdsPdf) | |
| 477 plotMDS(data, labels=data$samples$group, col=as.numeric(data$samples$group), | |
| 478 main="MDS Plot") | |
| 479 newEntry <- c("MDS Plot (.pdf)", "mds.pdf") | |
| 480 linkData <- rbind(linkData, newEntry) | |
| 481 invisible(dev.off()) | |
| 482 | |
|
9
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
483 # BCV Plot |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
484 png(bcvPng, width=600, height=600) |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
485 plotBCV(data, main="BCV Plot") |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
486 imageData <- rbind(imageData, c("BCV Plot", "bcv.png")) |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
487 invisible(dev.off()) |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
488 |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
489 pdf(bcvPdf) |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
490 plotBCV(data, main="BCV Plot") |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
491 newEntry <- c("BCV Plot (.pdf)", "bcv.pdf") |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
492 invisible(dev.off()) |
|
f1076bfb0ed1
- Fixed tool to actually make use of barcode location options
shian_su <registertonysu@gmail.com>
parents:
8
diff
changeset
|
493 |
| 2 | 494 if (workMode=="classic") { |
| 495 # Assess differential representation using classic exact testing methodology | |
| 496 # in edgeR | |
| 497 testData <- exactTest(data, pair=pairData) | |
| 498 | |
| 499 top <- topTags(testData, n=Inf) | |
| 500 topIDs <- top$table[(top$table$FDR < fdrThresh) & | |
| 501 (abs(top$table$logFC) > lfcThresh), 1] | |
| 7 | 502 |
| 2 | 503 write.table(top, file=topOut, row.names=FALSE, sep="\t") |
| 7 | 504 |
| 2 | 505 linkName <- paste0("Top Tags Table(", pairData[2], "-", pairData[1], |
| 506 ") (.tsv)") | |
| 507 linkAddr <- paste0("toptag(", pairData[2], "-", pairData[1], ").tsv") | |
| 508 linkData <- rbind(linkData, c(linkName, linkAddr)) | |
| 509 | |
| 7 | 510 upCount[1] <- sum(top$table$FDR < fdrThresh & top$table$logFC > lfcThresh) |
| 511 downCount[1] <- sum(top$table$FDR < fdrThresh & | |
| 512 top$table$logFC < -lfcThresh) | |
| 513 flatCount[1] <- sum(top$table$FDR > fdrThresh | | |
| 514 abs(top$table$logFC) < lfcThresh) | |
| 515 | |
| 516 | |
| 517 | |
| 2 | 518 # Select hairpins with FDR < 0.05 to highlight on plot |
| 519 png(smearPng, width=600, height=600) | |
| 520 plotTitle <- gsub(".", " ", | |
| 521 paste0("Smear Plot: ", pairData[2], "-", pairData[1]), | |
| 7 | 522 fixed=TRUE) |
| 2 | 523 plotSmear(testData, de.tags=topIDs, |
| 524 pch=20, cex=1.0, main=plotTitle) | |
| 7 | 525 abline(h=c(-1, 0, 1), col=c("dodgerblue", "yellow", "dodgerblue"), lty=2) |
| 2 | 526 imgName <- paste0("Smear Plot(", pairData[2], "-", pairData[1], ")") |
| 527 imgAddr <- paste0("smear(", pairData[2], "-", pairData[1],").png") | |
| 528 imageData <- rbind(imageData, c(imgName, imgAddr)) | |
| 529 invisible(dev.off()) | |
| 530 | |
| 531 pdf(smearPdf) | |
| 532 plotTitle <- gsub(".", " ", | |
| 533 paste0("Smear Plot: ", pairData[2], "-", pairData[1]), | |
| 7 | 534 fixed=TRUE) |
| 2 | 535 plotSmear(testData, de.tags=topIDs, |
| 536 pch=20, cex=1.0, main=plotTitle) | |
| 7 | 537 abline(h=c(-1, 0, 1), col=c("dodgerblue", "yellow", "dodgerblue"), lty=2) |
| 2 | 538 imgName <- paste0("Smear Plot(", pairData[2], "-", pairData[1], ") (.pdf)") |
| 539 imgAddr <- paste0("smear(", pairData[2], "-", pairData[1], ").pdf") | |
| 540 linkData <- rbind(linkData, c(imgName, imgAddr)) | |
| 541 invisible(dev.off()) | |
| 7 | 542 |
| 2 | 543 } else if (workMode=="glm") { |
| 544 # Generating design information | |
| 545 factors <- factor(data$sample$group) | |
| 546 design <- model.matrix(~0+factors) | |
| 547 | |
| 548 colnames(design) <- gsub("factors", "", colnames(design), fixed=TRUE) | |
| 549 | |
| 550 # Split up contrasts seperated by comma into a vector | |
| 551 contrastData <- unlist(strsplit(contrastData, split=",")) | |
| 7 | 552 |
| 2 | 553 for (i in 1:length(contrastData)) { |
| 554 # Generate contrasts information | |
| 555 contrasts <- makeContrasts(contrasts=contrastData[i], levels=design) | |
| 556 | |
| 557 # Fit negative bionomial GLM | |
| 7 | 558 fit <- glmFit(data, design) |
| 2 | 559 # Carry out Likelihood ratio test |
| 7 | 560 testData <- glmLRT(fit, contrast=contrasts) |
| 2 | 561 |
| 562 # Select hairpins with FDR < 0.05 to highlight on plot | |
| 563 top <- topTags(testData, n=Inf) | |
| 564 topIDs <- top$table[(top$table$FDR < fdrThresh) & | |
| 565 (abs(top$table$logFC) > lfcThresh), 1] | |
| 566 write.table(top, file=topOut[i], row.names=FALSE, sep="\t") | |
| 567 | |
| 568 linkName <- paste0("Top Tags Table(", contrastData[i], ") (.tsv)") | |
| 569 linkAddr <- paste0("toptag(", contrastData[i], ").tsv") | |
| 570 linkData <- rbind(linkData, c(linkName, linkAddr)) | |
| 571 | |
| 7 | 572 # Collect counts for differential representation |
| 573 upCount[i] <- sum(top$table$FDR < fdrThresh & top$table$logFC > lfcThresh) | |
| 574 downCount[i] <- sum(top$table$FDR < fdrThresh & | |
| 575 top$table$logFC < -lfcThresh) | |
| 576 flatCount[i] <- sum(top$table$FDR > fdrThresh | | |
| 577 abs(top$table$logFC) < lfcThresh) | |
| 578 | |
| 2 | 579 # Make a plot of logFC versus logCPM |
| 580 png(smearPng[i], height=600, width=600) | |
| 581 plotTitle <- paste("Smear Plot:", gsub(".", " ", contrastData[i], | |
| 582 fixed=TRUE)) | |
| 583 plotSmear(testData, de.tags=topIDs, pch=20, cex=0.8, main=plotTitle) | |
| 584 abline(h=c(-1, 0, 1), col=c("dodgerblue", "yellow", "dodgerblue"), lty=2) | |
| 585 | |
| 586 imgName <- paste0("Smear Plot(", contrastData[i], ")") | |
| 587 imgAddr <- paste0("smear(", contrastData[i], ").png") | |
| 588 imageData <- rbind(imageData, c(imgName, imgAddr)) | |
| 589 invisible(dev.off()) | |
| 590 | |
| 591 pdf(smearPdf[i]) | |
| 592 plotTitle <- paste("Smear Plot:", gsub(".", " ", contrastData[i], | |
| 593 fixed=TRUE)) | |
| 594 plotSmear(testData, de.tags=topIDs, pch=20, cex=0.8, main=plotTitle) | |
| 595 abline(h=c(-1, 0, 1), col=c("dodgerblue", "yellow", "dodgerblue"), lty=2) | |
| 596 | |
| 597 linkName <- paste0("Smear Plot(", contrastData[i], ") (.pdf)") | |
| 598 linkAddr <- paste0("smear(", contrastData[i], ").pdf") | |
| 599 linkData <- rbind(linkData, c(linkName, linkAddr)) | |
| 600 invisible(dev.off()) | |
| 601 | |
| 602 genes <- as.character(data$genes$Gene) | |
| 603 unq <- unique(genes) | |
| 604 unq <- unq[!is.na(unq)] | |
| 605 geneList <- list() | |
| 606 for (gene in unq) { | |
| 607 if (length(which(genes==gene)) >= hairpinReq) { | |
| 608 geneList[[gene]] <- which(genes==gene) | |
| 609 } | |
| 610 } | |
| 611 | |
| 612 if (wantRoast) { | |
| 613 # Input preparaton for roast | |
| 7 | 614 nrot <- 9999 |
| 2 | 615 set.seed(602214129) |
| 616 roastData <- mroast(data, index=geneList, design=design, | |
| 617 contrast=contrasts, nrot=nrot) | |
| 618 roastData <- cbind(GeneID=rownames(roastData), roastData) | |
| 619 write.table(roastData, file=roastOut[i], row.names=FALSE, sep="\t") | |
| 620 linkName <- paste0("Gene Level Analysis Table(", contrastData[i], | |
| 621 ") (.tsv)") | |
| 8 | 622 linkAddr <- paste0("gene_level(", contrastData[i], ").tsv") |
| 2 | 623 linkData <- rbind(linkData, c(linkName, linkAddr)) |
| 624 if (selectOpt=="rank") { | |
| 625 selectedGenes <- rownames(roastData)[selectVals] | |
| 626 } else { | |
| 627 selectedGenes <- selectVals | |
| 628 } | |
| 629 | |
| 630 if (packageVersion("limma")<"3.19.19") { | |
| 631 png(barcodePng[i], width=600, height=length(selectedGenes)*150) | |
| 632 } else { | |
| 633 png(barcodePng[i], width=600, height=length(selectedGenes)*300) | |
| 634 } | |
| 635 par(mfrow=c(length(selectedGenes), 1)) | |
| 636 for (gene in selectedGenes) { | |
| 637 barcodeplot(testData$table$logFC, index=geneList[[gene]], | |
| 638 main=paste("Barcode Plot for", gene, "(logFCs)", | |
| 639 gsub(".", " ", contrastData[i])), | |
| 640 labels=c("Positive logFC", "Negative logFC")) | |
| 641 } | |
| 642 imgName <- paste0("Barcode Plot(", contrastData[i], ")") | |
| 643 imgAddr <- paste0("barcode(", contrastData[i], ").png") | |
| 644 imageData <- rbind(imageData, c(imgName, imgAddr)) | |
| 645 dev.off() | |
| 646 if (packageVersion("limma")<"3.19.19") { | |
| 647 pdf(barcodePdf[i], width=8, height=2) | |
| 648 } else { | |
| 649 pdf(barcodePdf[i], width=8, height=4) | |
| 650 } | |
| 651 for (gene in selectedGenes) { | |
| 652 barcodeplot(testData$table$logFC, index=geneList[[gene]], | |
| 653 main=paste("Barcode Plot for", gene, "(logFCs)", | |
| 654 gsub(".", " ", contrastData[i])), | |
| 655 labels=c("Positive logFC", "Negative logFC")) | |
| 656 } | |
| 657 linkName <- paste0("Barcode Plot(", contrastData[i], ") (.pdf)") | |
| 658 linkAddr <- paste0("barcode(", contrastData[i], ").pdf") | |
| 659 linkData <- rbind(linkData, c(linkName, linkAddr)) | |
| 660 dev.off() | |
| 661 } | |
| 662 } | |
| 663 } | |
| 664 | |
| 8 | 665 # Generate data frame of the significant differences |
| 7 | 666 sigDiff <- data.frame(Up=upCount, Flat=flatCount, Down=downCount) |
| 667 if (workMode == "glm") { | |
| 668 row.names(sigDiff) <- contrastData | |
| 669 } else if (workMode == "classic") { | |
| 670 row.names(sigDiff) <- paste0(pairData[2], "-", pairData[1]) | |
| 671 } | |
| 672 | |
| 8 | 673 # Output table of summarised counts |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
674 ID <- rownames(data$counts) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
675 outputCounts <- cbind(ID, data$counts) |
| 8 | 676 write.table(outputCounts, file=countsOut, row.names=FALSE, sep="\t", |
| 677 quote=FALSE) | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
678 linkName <- "Counts table (.tsv)" |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
679 linkAddr <- "counts.tsv" |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
680 linkData <- rbind(linkData, c(linkName, linkAddr)) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
681 |
| 8 | 682 # Record session info |
| 683 writeLines(capture.output(sessionInfo()), sessionOut) | |
| 684 linkData <- rbind(linkData, c("Session Info", "session_info.txt")) | |
| 685 | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
686 # Record ending time and calculate total run time |
| 2 | 687 timeEnd <- as.character(Sys.time()) |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
688 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
|
689 timeTaken <- gsub("Time difference of ", "", timeTaken, fixed=TRUE) |
| 2 | 690 ################################################################################ |
| 691 ### HTML Generation | |
| 692 ################################################################################ | |
| 693 # Clear file | |
| 694 cat("", file=htmlPath) | |
| 695 | |
| 696 cata("<html>\n") | |
| 697 HtmlHead("EdgeR Output") | |
| 698 | |
| 699 cata("<body>\n") | |
| 700 cata("<h3>EdgeR Analysis Output:</h3>\n") | |
| 701 cata("<h4>Input Summary:</h4>\n") | |
| 702 if (inputType=="fastq") { | |
| 703 cata("<ul>\n") | |
| 704 ListItem(hpReadout[1]) | |
| 705 ListItem(hpReadout[2]) | |
| 706 cata("</ul>\n") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
707 cata(hpReadout[3], "<br />\n") |
| 2 | 708 cata("<ul>\n") |
| 709 ListItem(hpReadout[4]) | |
| 710 ListItem(hpReadout[7]) | |
| 711 cata("</ul>\n") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
712 cata(hpReadout[8:11], sep="<br />\n") |
| 2 | 713 cata("<br />\n") |
| 714 cata("<b>Please check that read percentages are consistent with ") | |
| 715 cata("expectations.</b><br >\n") | |
| 716 } else if (inputType=="counts") { | |
| 717 cata("<ul>\n") | |
| 718 ListItem("Number of Samples: ", ncol(data$counts)) | |
| 719 ListItem("Number of Hairpins: ", countsRows) | |
| 720 ListItem("Number of annotations provided: ", annoRows) | |
| 721 ListItem("Number of annotations matched to hairpin: ", annoMatched) | |
| 722 cata("</ul>\n") | |
| 723 } | |
| 724 | |
| 725 cata("The estimated common biological coefficient of variation (BCV) is: ", | |
| 726 commonBCV, "<br />\n") | |
| 727 | |
| 728 cata("<h4>Output:</h4>\n") | |
| 7 | 729 cata("PDF copies of JPEGS available in 'Plots' section.<br />\n") |
| 2 | 730 for (i in 1:nrow(imageData)) { |
| 731 if (grepl("barcode", imageData$Link[i])) { | |
| 732 if (packageVersion("limma")<"3.19.19") { | |
| 733 HtmlImage(imageData$Link[i], imageData$Label[i], | |
| 734 height=length(selectedGenes)*150) | |
| 735 } else { | |
| 736 HtmlImage(imageData$Link[i], imageData$Label[i], | |
| 737 height=length(selectedGenes)*300) | |
| 738 } | |
| 739 } else { | |
| 740 HtmlImage(imageData$Link[i], imageData$Label[i]) | |
| 741 } | |
| 742 } | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
743 cata("<br />\n") |
| 2 | 744 |
| 7 | 745 cata("<h4>Differential Representation Counts:</h4>\n") |
| 746 | |
| 747 cata("<table border=\"1\" cellpadding=\"4\">\n") | |
| 748 cata("<tr>\n") | |
| 749 TableItem() | |
| 750 for (i in colnames(sigDiff)) { | |
| 751 TableHeadItem(i) | |
| 752 } | |
| 753 cata("</tr>\n") | |
| 754 for (i in 1:nrow(sigDiff)) { | |
| 755 cata("<tr>\n") | |
| 756 TableHeadItem(unmake.names(row.names(sigDiff)[i])) | |
| 757 for (j in 1:ncol(sigDiff)) { | |
| 758 TableItem(as.character(sigDiff[i, j])) | |
| 759 } | |
| 760 cata("</tr>\n") | |
| 761 } | |
| 762 cata("</table>") | |
| 763 | |
| 2 | 764 cata("<h4>Plots:</h4>\n") |
| 765 for (i in 1:nrow(linkData)) { | |
| 8 | 766 if (grepl(".pdf", linkData$Link[i])) { |
| 2 | 767 HtmlLink(linkData$Link[i], linkData$Label[i]) |
| 768 } | |
| 769 } | |
| 770 | |
| 771 cata("<h4>Tables:</h4>\n") | |
| 772 for (i in 1:nrow(linkData)) { | |
| 773 if (grepl(".tsv", linkData$Link[i])) { | |
| 774 HtmlLink(linkData$Link[i], linkData$Label[i]) | |
| 775 } | |
| 776 } | |
| 777 | |
| 7 | 778 cata("<p>Alt-click links to download file.</p>\n") |
| 779 cata("<p>Click floppy disc icon associated history item to download ") | |
| 780 cata("all files.</p>\n") | |
| 781 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
|
782 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
783 cata("<h4>Additional Information:</h4>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
784 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
785 if (inputType == "fastq") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
786 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
|
787 } else if (inputType == "counts") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
788 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
|
789 } |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
790 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
791 if (cpmReq!=0 && sampleReq!=0) { |
| 8 | 792 tempStr <- paste("Hairpins without more than", cpmReq, |
| 7 | 793 "CPM in at least", sampleReq, "samples are insignificant", |
| 794 "and filtered out.") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
795 ListItem(tempStr) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
796 filterProp <- round(filteredCount/preFilterCount*100, digits=2) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
797 tempStr <- paste0(filteredCount, " of ", preFilterCount," (", filterProp, |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
798 "%) 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
|
799 ListItem(tempStr) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
800 } |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
801 |
|
10
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
802 if (exists("filteredSamples")) { |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
803 tempStr <- paste("The following samples were filtered out for having zero", |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
804 "library size: ", filteredSamples) |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
805 ListItem(tempStr) |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
806 } |
|
8923d4ea858b
- Added check for zero library size, will now filter out zero library size
shian_su <registertonysu@gmail.com>
parents:
9
diff
changeset
|
807 |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
808 if (workMode == "classic") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
809 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
|
810 } else if (workMode == "glm") { |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
811 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
|
812 } |
|
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 cit <- character() |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
815 link <-character() |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
816 link[1] <- paste0("<a href=\"", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
817 "http://www.bioconductor.org/packages/release/bioc/", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
818 "vignettes/limma/inst/doc/usersguide.pdf", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
819 "\">", "limma User's Guide", "</a>.") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
820 link[2] <- paste0("<a href=\"", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
821 "http://www.bioconductor.org/packages/release/bioc/", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
822 "vignettes/edgeR/inst/doc/edgeRUsersGuide.pdf", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
823 "\">", "edgeR User's Guide", "</a>") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
824 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
825 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
|
826 "edgeR: a Bioconductor package for differential", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
827 "expression analysis of digital gene expression", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
828 "data. Bioinformatics 26, 139-140") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
829 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
|
830 "for assessing differences in tag abundance. Bioinformatics", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
831 "23, 2881-2887") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
832 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
|
833 "negative binomial dispersion, with applications to SAGE data.", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
834 "Biostatistics, 9, 321-332") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
835 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
836 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
|
837 "expression analysis of multifactor RNA-Seq experiments with", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
838 "respect to biological variation. Nucleic Acids Research 40,", |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
839 "4288-4297") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
840 |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
841 cata("<h4>Citations</h4>") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
842 cata("<ol>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
843 ListItem(cit[1]) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
844 ListItem(cit[2]) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
845 ListItem(cit[3]) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
846 ListItem(cit[4]) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
847 cata("</ol>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
848 |
| 8 | 849 cata("<p>Report problems to: su.s@wehi.edu.au</p>\n") |
| 850 | |
| 851 for (i in 1:nrow(linkData)) { | |
| 852 if (grepl("session_info", linkData$Link[i])) { | |
| 853 HtmlLink(linkData$Link[i], linkData$Label[i]) | |
| 854 } | |
| 855 } | |
| 856 | |
| 2 | 857 cata("<table border=\"0\">\n") |
| 858 cata("<tr>\n") | |
| 859 TableItem("Task started at:"); TableItem(timeStart) | |
| 860 cata("</tr>\n") | |
| 861 cata("<tr>\n") | |
| 862 TableItem("Task ended at:"); TableItem(timeEnd) | |
| 863 cata("</tr>\n") | |
|
6
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
864 cata("<tr>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
865 TableItem("Task run time:"); TableItem(timeTaken) |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
866 cata("<tr>\n") |
|
3d04308a99f9
- Added differentially expressed hairpin count output
shian_su <registertonysu@gmail.com>
parents:
4
diff
changeset
|
867 cata("</table>\n") |
| 2 | 868 |
| 869 cata("</body>\n") | |
| 870 cata("</html>") |