cont cutadapt pe report (still incomplete)

dge_workflow/deprecated/cutadapt_summary_pe.R

@@ -78,24 +78,24 @@ readSummaryStats <- function(logFile){
     # DEBUG logFile=logDataFiles[1]
   data.frame(
     Run=sub("^([^.]*).*", "\\1", basename(logFile)) %>% tidySamples,
-    No_of_processed_Reads=(paste("grep -F 'Total read' ", logFile )  %>% pipe() %>% readLines() %>% strsplit( "[^0-9]+") %>% unlist() %>% as.numeric())[2],
-    No_of_processed_Bases=(paste("grep -F 'Total basepairs processed' ", 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]
+    No_of_processed_Reads=paste("grep -F 'Total read pairs processed' ", logFile )  %>% pipe() %>% readLines() %>% str_split(fixed(":")) %>% unlist() %>% {.[2]} %>% str_trim %>% str_replace(",", "") %>% as.numeric(),
+    No_of_processed_Bases=paste("grep -F 'Total basepairs processed' ", logFile )  %>% pipe() %>% readLines() %>% str_replace_all(",", "") %>% 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() %>%  str_replace_all(",", "") %>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric()%>% {.[3]},
+    # 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' ", logFile )  %>% pipe() %>% readLines()  %>%  str_replace_all(",", "")%>% strsplit( "[^0-9\\.]+") %>% unlist() %>% as.numeric()%>% {.[3]}
   )
 }
 trimmingStats <- logDataFiles %>% map_df(readSummaryStats)
 
-#+ results = 'asis'
-trimmingStats %>% head() %>% kable()
+trimmingStats %>% table_browser()
 
 write_tsv(trimmingStats, file="cutadapt_summary.trimmingStats.txt")
 #'  [Trimming Statistics](cutadapt_summary.trimmingStats.txt)
 
-myfun <- function(x) x/100
-trimmingStats <- mutate_each(trimmingStats, funs(myfun), Trimmed_Reads, Quality_trimmed, Trimmed_Bases, Too_short_Reads)
+# myfun <- function(x) x/100
+# trimmingStats <- mutate_each(trimmingStats, funs(myfun), Trimmed_Reads, Quality_trimmed, Trimmed_Bases, Too_short_Reads)
 
 #+ fig.height=nrow(trimmingStats)/3, fig.width=12
 require(scales)