routine sync

routes/rna-star-groups-dge.sh

@@ -138,7 +138,7 @@ sleep 3
 #########################
 
 
-echo -e "\n ========== test R ========== \n"
+echo -e "\n ========== test R environment ========== \n"
 
 # load relevant modules
 module unload java
@@ -154,10 +154,9 @@ echo " * Rscript: $(readlink -f $(which Rscript)) "
 echo " * Rscript version: $(Rscript --version 2>&1) "
 echo
 
-# test R installation and setting
-bash_cmd="Rscript --vanilla ${code_dir}/scripts/test-packages.R"
-echo "CMD: $bash_cmd"
-($bash_cmd)
+Rscript --vanilla ${code_dir}/scripts/test-package.R optparse
+Rscript --vanilla ${code_dir}/scripts/test-package.R mnormt
+Rscript --vanilla ${code_dir}/scripts/test-package.R limma
 
 sleep 3
 

scripts/dge-deseq2.R

@@ -28,6 +28,9 @@ genes_gtf = args[1]
 counts_table_file = args[2]
 groups_table_file = args[3]
 
+# check for arguments
+if (length(args) < 3) stop("not enough arguments provided")
+
 # check that input files exist
 if (!file.exists(counts_table_file)) stop("file does not exist: ", counts_table_file)
 if (!file.exists(groups_table_file)) stop("file does not exist: ", groups_table_file)

scripts/load-install-packages.R

@@ -12,30 +12,17 @@ load_install_packages = function(package_list) {
     # create local user library path (not present by default)
     dir.create(path = Sys.getenv("R_LIBS_USER"), showWarnings = FALSE, recursive = TRUE)
 
-    # try to install from CRAN if package is not already installed
+    # try to install using biocLite() (installs both Bioconductor and CRAN packages)
     # use require() instead of installed.packages() to check that package is both installed and usable
-
-    # biocLite can install CRAN as well
-    # if (!require(p, character.only = TRUE, quietly = TRUE)) {
-    #   message("try installing package from CRAN: ", p)
-    #   # install to local user library path (destination needs to be specified the first time)
-    #   install.packages(p, lib = Sys.getenv("R_LIBS_USER"), repos = "https://cran.rstudio.com/", quiet = TRUE)
-    # }
-
-    # try to install from Bioconductor if package is not already installed
-    # biocLite() installs Bioconductor and CRAN packages
     if (!require(p, character.only = TRUE, quietly = TRUE)) {
-      # message("package not available on CRAN: ", p)
-      # message("try installing package from Bioconductor: ", p)
-      message("try installing package: ", p)
+      message("installing package: ", p)
       source("https://bioconductor.org/biocLite.R")
       biocLite(p, suppressUpdates = TRUE, lib = Sys.getenv("R_LIBS_USER"))
     }
 
     # package still not available (after CRAN and Bioconductor)
     if (!require(p, character.only = TRUE, quietly = TRUE)) {
-      # message("package not available on Bioconductor: ", p)
-      message("package not available: ", p)
+      message("package not available on CRAN or Bioconductor: ", p)
     }
 
     # load package

scripts/test-package.R

@@ -0,0 +1,28 @@
+#!/usr/bin/env Rscript
+
+
+##
+## Try loading an R package. Install to local user library if not present.
+##
+## usage: Rscript --vanilla test-package.R package
+##
+
+
+# get scripts directory (directory of this file) and load relevant functions
+args_all = commandArgs(trailingOnly = FALSE)
+scripts_dir = normalizePath(dirname(sub("^--file=", "", args_all[grep("^--file=", args_all)])))
+source(paste0(scripts_dir, "/load-install-packages.R"))
+
+# relevent arguments
+args = commandArgs(trailingOnly = TRUE)
+package_name = args[1]
+
+# check for arguments
+if (length(args) < 1) stop("not enough arguments provided")
+
+# load relevant packages
+load_install_packages(package_name)
+
+
+
+# end

scripts/test-packages.R

@@ -2,7 +2,7 @@
 
 
 ##
-## Try loading (installing if not available) a few R packages.
+## Try loading some common R packages. Install to local user library if not present.
 ##
 ## usage: Rscript --vanilla test-packages.R
 ##
@@ -13,7 +13,7 @@ args_all = commandArgs(trailingOnly = FALSE)
 scripts_dir = normalizePath(dirname(sub("^--file=", "", args_all[grep("^--file=", args_all)])))
 source(paste0(scripts_dir, "/load-install-packages.R"))
 
-# load some packages (some will probably need to be installed)
+# load some packages (some are rare and will probably need to be installed)
 # used by GATK: ggplot2, gplots, reshape, gsalib
 test_packages = c("optparse", "bitops", "mnormt", "reshape", "gplots", "ggplot2", "gsalib", "dplyr", "limma")
 load_install_packages(test_packages)

segments/annot-annovar.sh

@@ -307,7 +307,7 @@ echo "total muts: $total_muts"
 coding_muts=$(cat "$annovar_combined" | grep -v 'refGene' | grep 'exon' | wc -l)
 echo "coding muts: $coding_muts"
 
-nonsyn_muts=$(cat "$annovar_combined" | grep -v 'refGene' | grep -E 'nonsynonymous|stopgain|frameshift' | wc -l)
+nonsyn_muts=$(cat "$annovar_combined" | grep -v 'refGene' | grep -E 'nonsynonymous|stopgain|stoploss|frameshift' | wc -l)
 echo "nonsynonymous muts: $coding_muts"
 
 # header for summary file
@@ -329,7 +329,10 @@ cat ${summary_dir}/*.${segment_name}.csv | LC_ALL=C sort -t ',' -k1,1 | uniq > "
 
 # all mutations
 combine_all_cmd="
-cat ${annovar_dir}/*.combined.txt | LC_ALL=C sort -k1,1 -k2,2 | uniq > ${annovar_dir}.all.txt
+cat ${annovar_dir}/*.combined.txt \
+| LC_ALL=C sort -k1,1 -k2,2 \
+| uniq \
+> ${annovar_dir}.all.txt
 "
 echo -e "\n CMD: $combine_all_cmd \n"
 eval "$combine_all_cmd"
@@ -338,7 +341,9 @@ sleep 1
 
 # coding mutations
 combine_coding_cmd="
-cat ${annovar_dir}.all.txt | grep -E 'refGene|exon|splicing' > ${annovar_dir}.coding.txt
+cat ${annovar_dir}.all.txt \
+| grep -E 'refGene|exon|splicing' \
+> ${annovar_dir}.coding.txt
 "
 echo -e "\n CMD: $combine_coding_cmd \n"
 eval "$combine_coding_cmd"
@@ -347,7 +352,9 @@ sleep 1
 
 # consequence mutations
 combine_nonsyn_cmd="
-cat ${annovar_dir}.all.txt | grep -E 'refGene|nonsynonymous|stopgain|splicing|frameshift' > ${annovar_dir}.nonsyn.txt
+cat ${annovar_dir}.all.txt \
+| grep -E 'refGene|splicing|nonsynonymous|stopgain|stoploss|frameshift' \
+> ${annovar_dir}.nonsyn.txt
 "
 echo -e "\n CMD: $combine_nonsyn_cmd \n"
 eval "$combine_nonsyn_cmd"

segments/fastq-clean.sh

@@ -12,7 +12,7 @@ echo -e "\n ========== SEGMENT: $segment_name ========== \n" >&2
 # check for correct number of arguments
 if [ ! $# == 2 ] ; then
 	echo -e "\n $script_name ERROR: WRONG NUMBER OF ARGUMENTS SUPPLIED \n" >&2
-	echo -e "\n USAGE: $script_name [project dir] [sample name] \n" >&2
+	echo -e "\n USAGE: $script_name project_dir sample_name \n" >&2
 	exit 1
 fi