Mercurial > repos > mingchen0919 > rmarkdown_deseq2
diff DESeq_results_render.R @ 6:2f8ddef8d545 draft
update deseq2
author | mingchen0919 |
---|---|
date | Tue, 07 Nov 2017 13:50:32 -0500 |
parents | 7231d7e8d3ed |
children |
line wrap: on
line diff
--- a/DESeq_results_render.R Tue Aug 08 15:06:40 2017 -0400 +++ b/DESeq_results_render.R Tue Nov 07 13:50:32 2017 -0500 @@ -1,120 +1,108 @@ -##======= Handle arguments from command line ======== -# setup R error handline to go to stderr -options(show.error.messages=FALSE, - error=function(){ - cat(geterrmessage(), file=stderr()) - quit("no", 1, F) - }) - -# we need that to not crash galaxy with an UTF8 error on German LC settings. -loc = Sys.setlocale("LC_MESSAGES", "en_US.UTF-8") - -# suppress warning -options(warn = -1) - -options(stringsAsFactors=FALSE, useFancyQuotes=FALSE) -args = commandArgs(trailingOnly=TRUE) - -suppressPackageStartupMessages({ - library(getopt) - library(tools) -}) +library(getopt) +library(rmarkdown) +library(htmltools) +library(dplyr) +library(DESeq2) +library(pheatmap) +library(genefilter) +library(DT) +library(stringi) +library(RColorBrewer) +library(ggplot2) -# column 1: the long flag name -# column 2: the short flag alias. A SINGLE character string -# column 3: argument mask -# 0: no argument -# 1: argument required -# 2: argument is optional -# column 4: date type to which the flag's argument shall be cast. -# possible values: logical, integer, double, complex, character. -spec_list=list() - -##------- 1. input data --------------------- -spec_list$ECHO = c('echo', 'e', '1', 'character') -spec_list$DESEQ_WORKSPACE = c('deseq_workspace', 'w', '1', 'character') -spec_list$SAMPLE_TABLE = c('sample_table', 's', '1', 'character') -spec_list$CONTRAST_GROUP = c('contrast_group', 'c', '1', 'character') -spec_list$TREATMENT_LEVEL = c('treatment_level', 't', '1', 'character') -spec_list$CONDITION_LEVEL = c('condition_level', 'k', '1', 'character') -spec_list$CLUSTERING_GROUPS = c('clustering_groups', 'm', '1', 'character') - -##--------2. output report and report site directory -------------- -spec_list$OUTPUT_HTML = c('deseq_results_html', 'o', '1', 'character') -spec_list$OUTPUT_DIR = c('deseq_results_dir', 'd', '1', 'character') - -##--------3. Rmd templates sitting in the tool directory ---------- - -spec_list$DESEQ_VISUALIZATION_RMD = c('deseq_results_rmd', 'D', '1', 'character') - +##============ Sink warnings and errors to a file ============== +## use the sink() function to wrap all code within it. +##============================================================== +zz = file('warnings_and_errors.txt') +sink(zz) +sink(zz, type = 'message') + ##---------below is the code for rendering .Rmd templates----- + + ##=============STEP 1: handle command line arguments========== + ## + ##============================================================ + # column 1: the long flag name + # column 2: the short flag alias. A SINGLE character string + # column 3: argument mask + # 0: no argument + # 1: argument required + # 2: argument is optional + # column 4: date type to which the flag's argument shall be cast. + # possible values: logical, integer, double, complex, character. + #------------------------------------------------------------- + #++++++++++++++++++++ Best practice ++++++++++++++++++++++++++ + # 1. short flag alias should match the flag in the command section in the XML file. + # 2. long flag name can be any legal R variable names + # 3. two names in args_list can have common string but one name should not be a part of another name. + # for example, one name is "ECHO", if another name is "ECHO_XXX", it will cause problems. + #+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ + args_list=list() + ##------- 1. input data --------------------- + args_list$ECHO = c('echo', 'e', '1', 'character') + args_list$DESEQ_WORKSPACE = c('deseq_workspace', 'W', '1', 'character') + args_list$CONTRAST_FACTOR = c('contrast_factor', 'C', '1', 'character') + args_list$TREATMENT_LEVEL = c('treatment_level', 'T', '1', 'character') + args_list$CONDITION_LEVEL = c('condition_level', 'K', '1', 'character') + args_list$CLUSTERING_FACTORS = c('clustering_factors', 'M', '1', 'character') + ##--------2. output report and outputs -------------- + args_list$REPORT_HTML = c('report_html', 'r', '1', 'character') + args_list$REPORT_DIR = c('report_dir', 'd', '1', 'character') + args_list$SINK_MESSAGE = c('sink_message', 's', '1', 'character') + args_list$DESEQ_RESULTS = c('deseq_results', 'R', '1', 'character') + ##--------3. .Rmd templates in the tool directory ---------- + args_list$deseq_results_RMD = c('deseq_results_rmd', 't', '1', 'character') + ##----------------------------------------------------------- + opt = getopt(t(as.data.frame(args_list))) -##------------------------------------------------------------------ - -spec = t(as.data.frame(spec_list)) -opt = getopt(spec) -# arguments are accessed by long flag name (the first column in the spec matrix) -# NOT by element name in the spec_list -# example: opt$help, opt$expression_file -##====== End of arguments handling ========== + + ##=======STEP 2: create report directory (optional)========== + ## + ##=========================================================== + dir.create(opt$report_dir) + + ##=STEP 3: replace placeholders in .Rmd with argument values= + ## + ##=========================================================== + #++ need to replace placeholders with args values one by one+ + readLines(opt$deseq_results_rmd) %>% + (function(x) { + gsub('ECHO', opt$echo, x) + }) %>% + (function(x) { + gsub('DESEQ_WORKSPACE', opt$deseq_workspace, x) + }) %>% + (function(x) { + gsub('CONTRAST_FACTOR', opt$contrast_factor, x) + }) %>% + (function(x) { + gsub('TREATMENT_LEVEL', opt$treatment_level, x) + }) %>% + (function(x) { + gsub('CONDITION_LEVEL', opt$condition_level, x) + }) %>% + (function(x) { + gsub('CLUSTERING_FACTORS', opt$clustering_factors, x) + }) %>% + (function(x) { + gsub('REPORT_DIR', opt$report_dir, x) + }) %>% + (function(x) { + gsub('DESEQ_RESULTS', opt$deseq_results, x) + }) %>% + (function(x) { + fileConn = file('deseq_results.Rmd') + writeLines(x, con=fileConn) + close(fileConn) + }) + -#------ Load libraries --------- -library(rmarkdown) -library(plyr) -library(stringr) -library(dplyr) -library(highcharter) -library(DT) -library(reshape2) -# library(Kmisc) -library(plotly) -library(formattable) -library(htmltools) - - -#----- 1. create the report directory ------------------------ -system(paste0('mkdir -p ', opt$deseq_results_dir)) - - -#----- 2. generate Rmd files with Rmd templates -------------- -# a. templates without placeholder variables: -# copy templates from tool directory to the working directory. -# b. templates with placeholder variables: -# substitute variables with user input values and place them in the working directory. + ##=============STEP 4: render .Rmd templates================= + ## + ##=========================================================== + render('deseq_results.Rmd', output_file = opt$report_html) -#----- 01 DESeq_results.Rmd ----------------------- -readLines(opt$deseq_results_rmd) %>% - (function(x) { - gsub('ECHO', opt$echo, x) - }) %>% - (function(x) { - gsub('DESEQ_WORKSPACE', opt$deseq_workspace, x) - }) %>% - (function(x) { - gsub('CONTRAST_GROUP', opt$contrast_group, x) - }) %>% - (function(x) { - gsub('TREATMENT_LEVEL', opt$treatment_level, x) - }) %>% - (function(x) { - gsub('CONDITION_LEVEL', opt$condition_level, x) - }) %>% - (function(x) { - gsub('CLUSTERING_GROUPS', opt$clustering_groups, x) - }) %>% - (function(x) { - gsub('OUTPUT_DIR', opt$deseq_results_dir, x) - }) %>% - (function(x) { - fileConn = file('DESeq_results.Rmd') - writeLines(x, con=fileConn) - close(fileConn) - }) - - -#------ 3. render all Rmd files -------- -render('DESeq_results.Rmd', output_file = opt$deseq_results_html) - - -#-------4. manipulate outputs ----------------------------- + ##--------end of code rendering .Rmd templates---------------- +sink() +##=========== End of sinking output============================= \ No newline at end of file