new variable names

dge_workflow/cutadapt_summary.R

@@ -6,6 +6,7 @@
 
 devtools::source_url("https://dl.dropboxusercontent.com/u/113630701/datautils/R/core_commons.R")
 devtools::source_url("https://dl.dropboxusercontent.com/u/113630701/datautils/R/ggplot_commons.R")
+require(knitr)
 
 ## can we access variables from the parent spin.R process?
 #echo("rscript is ", r_script)
@@ -17,7 +18,6 @@ argv = commandArgs(TRUE)
 
 if(length(argv) != 1){
   stop("Usage: First_try.R <directory with cutadapt log files>")
-  echo
 }
 
 baseDir=argv[1]
@@ -47,14 +47,13 @@ echo(info3)
 parameters=readLines(pipe( paste("grep -F 'Command line parameters' ", logDataFiles[1]) ))
 echo(parameters)
 
-
-echo("Some explanation:")
+#' #### Some explanations:
 if (grepl("-a", parameters) ==TRUE)
-  echo("-a indicates that the following is a 3' adapter.")
+  echo("-a indicates that the following is a 3' end adapter.")
 if (grepl("-g", parameters) ==TRUE)
-  echo("-g indicates that the following is a 5' adapter.")
+  echo("-g indicates that the following is a 5' end adapter.")
 if (grepl("-b", parameters) ==TRUE)
-  echo("-b indicates that the adapter is 3' or 5' (both possible).")
+  echo("-b indicates that the adapter is located at the 3' or 5' end (both possible).")
 if (grepl("-m", parameters) ==TRUE)
   echo("Reads shorter than -m bases are thrown away.")
 if (grepl("-q", parameters) ==TRUE)
@@ -68,19 +67,19 @@ if (grepl("-O", parameters) ==TRUE)
 if (grepl("-N", parameters) ==TRUE)
   echo("Wildcard characters in the adapter are enabled by -N.")
 
-echo("For more detailed information on cutadapt go to https://cutadapt.readthedocs.org/en/latest/index.html")
+#' #### For more detailed information on cutadapt go to https://cutadapt.readthedocs.org/en/latest/index.html
 
 
 #' ## Trimming Overview
 genTable1 <- function(logFile){
   data.frame(
-    run=sub("^([^.]*).*", "\\1", basename(logFile)),
-    num_proReads=(paste("grep -F 'Processed reads' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9]+") %>% unlist() %>% as.numeric())[2],
-    num_proBases=(paste("grep -F 'Processed bases' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9]+") %>% unlist() %>% as.numeric())[2],
-    trim_reads=(paste("grep -F 'Trimmed reads' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric())[3],
-    qual_trimmed=(paste("grep -F 'Quality-trimmed' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric())[4],
-    trim_bases=(paste("grep -F 'Trimmed bases' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric())[4],
-    too_short=(paste("grep -F 'Too short reads' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric())[3]
+    Run=sub("^([^.]*).*", "\\1", basename(logFile)),
+    No_of_processed_Reads=(paste("grep -F 'Processed reads' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9]+") %>% unlist() %>% as.numeric())[2],
+    No_of_processed_Bases=(paste("grep -F 'Processed bases' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9]+") %>% unlist() %>% as.numeric())[2],
+    Trimmed_Reads=(paste("grep -F 'Trimmed reads' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric())[3],
+    Quality-trimmed=(paste("grep -F 'Quality-trimmed' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric())[4],
+    Trimmed_Bases=(paste("grep -F 'Trimmed bases' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric())[4],
+    Too_short_Reads=(paste("grep -F 'Too short reads' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric())[3]
   )
 }
 trimmingStats <- logDataFiles %>% ldply(genTable1)
@@ -95,11 +94,11 @@ write.delim(trimmingStats, file="cutadapt_summary.trimmingStats.txt")
 genTable2 <- function(logFile){
   #browser()
   data.frame(
-    run=sub("^([^.]*).*", "\\1", basename(logFile)),
-    adapter=(paste("grep -F '=== Adapter ' ", logFile )  %>% pipe() %>% readLines() %>% str_split_fixed("'", 3))[,2],
-    trimmed=(paste("grep -F '; Trimmed: ' ", logFile )  %>% pipe() %>% readLines() %>% str_split_fixed( "[^0-9]+", 6))[,5] %>% as.numeric(),
-    overlapped5=(paste("grep -F 'overlapped the 5' ", logFile )  %>% pipe() %>% readLines() %>% str_split_fixed( "[^0-9]+", 2) )[,1] %>% as.numeric(),
-    overlapped3=(paste("grep -F 'overlapped the 3' ", logFile )  %>% pipe() %>% readLines() %>% str_split_fixed( "[^0-9]+", 2) )[,1] %>% as.numeric()
+    Run=sub("^([^.]*).*", "\\1", basename(logFile)),
+    Adapter=(paste("grep -F '=== Adapter ' ", logFile )  %>% pipe() %>% readLines() %>% str_split_fixed("'", 3))[,2],
+    Trimmed=(paste("grep -F '; Trimmed: ' ", logFile )  %>% pipe() %>% readLines() %>% str_split_fixed( "[^0-9]+", 6))[,5] %>% as.numeric(),
+    Overlapped_at_5prime=(paste("grep -F 'overlapped the 5' ", logFile )  %>% pipe() %>% readLines() %>% str_split_fixed( "[^0-9]+", 2) )[,1] %>% as.numeric(),
+    Overlapped_at_3prime=(paste("grep -F 'overlapped the 3' ", logFile )  %>% pipe() %>% readLines() %>% str_split_fixed( "[^0-9]+", 2) )[,1] %>% as.numeric()
   )
 }
 adapterTrimmingStats <- logDataFiles %>% ldply(genTable2)
@@ -112,8 +111,13 @@ write.delim(adapterTrimmingStats, file="cutadapt_summary.adapterTrimmingStats.tx
 #'  [Adapter Statistics](cutadapt_summary.adapterTrimmingStats.txt)
 
 #+ fig.height=10, fig.width=10
-ggplot(trimmingStats, aes(run, num_proReads)) + geom_bar(stat='identity') + coord_flip()
-
-ggplot(adapterTrimmingStats, aes(run, trimmed)) + geom_bar(stat='identity') + facet_wrap(~adapter) + coord_flip()
-ggplot(adapterTrimmingStats, aes(run, overlapped5)) + geom_bar(stat='identity') + facet_wrap(~adapter) + coord_flip()
-ggplot(adapterTrimmingStats, aes(run, overlapped3)) + geom_bar(stat='identity') + facet_wrap(~adapter) + coord_flip()
+ggplot(trimmingStats, aes(Run, No_of_processed_Reads)) + geom_bar(stat='identity') + coord_flip()
+ggplot(trimmingStats, aes(Run, No_of_processed_Bases)) + geom_bar(stat='identity') + coord_flip()
+ggplot(trimmingStats, aes(Run, Trimmed_Reads)) + geom_bar(stat='identity') + coord_flip()
+ggplot(trimmingStats, aes(Run, Quality-trimmed)) + geom_bar(stat='identity') + coord_flip()
+ggplot(trimmingStats, aes(Run, Trimmed_Bases)) + geom_bar(stat='identity') + coord_flip()
+ggplot(trimmingStats, aes(Run, Too_short_Reads)) + geom_bar(stat='identity') + coord_flip()
+
+ggplot(adapterTrimmingStats, aes(Run, Trimmed)) + geom_bar(stat='identity') + facet_wrap(~Adapter) + coord_flip()
+ggplot(adapterTrimmingStats, aes(Run, Overlapped_at_5prime)) + geom_bar(stat='identity') + facet_wrap(~Adapter) + coord_flip()
+ggplot(adapterTrimmingStats, aes(Run, Overlapped_at_3prime)) + geom_bar(stat='identity') + facet_wrap(~Adapter) + coord_flip()