integrated cutadapt log parser into workflow

dge_workflow/dge_master_template.sh

@@ -11,7 +11,7 @@ export PATH=$DGE_HOME:$DGE_HOME/../misc/:$PATH
 ########################################################################################################################
 ### QC
 
-dge_fastqc *fastq.gz &
+dge_fastqc $(ls *fastq.gz) &
 
 
 ########################################################################################################################
@@ -21,7 +21,7 @@ mcdir $baseDir/trimmed
 
 dge_cutadapt $(ls $baseDir/treps_pooled/*fastq.gz) 2>&1 | tee cutadapt.log
 
-dge_fastqc $(ls *fastq.gz)
+dge_fastqc $(ls *fastq.gz) &
 
 
 ########################################################################################################################

dge_workflow/dge_utils.sh

@@ -82,6 +82,9 @@ for fastqFile in $* ; do
 done
 
 wait4jobs .cajobs
+
+spin.R $DGE_HOME/cutadapt_summary.R .
+
 ziprm cutadapt_logs ${project}__ca__*.log
 
 

dge_workflow/fastqc_summary.R

@@ -49,7 +49,10 @@ readCount <- function(statsFile){
 readCounts <- fastqDataFiles %>% ldply(readCount) %>% print_head()
 
 require.auto(scales)
-ggplot(readCounts, aes(run, num_reads)) + geom_bar(stat="identity") + coord_flip() + scale_y_continuous(labels=comma)
+gg <- ggplot(readCounts, aes(run, num_reads)) + geom_bar(stat="identity") + coord_flip() + scale_y_continuous(labels=comma) + ggtitle("read counts")
+
+#+ results='asis'
+gg %>%  ggsave2(width=15, height = round(nrow(readCounts)/4), limitsize=FALSE) %>% paste0("<img src='", ., "'><br>") %>% cat()
 
 
 ### Create a faill/pass matrix
@@ -68,10 +71,14 @@ readSummary <- function(statsFile){
 qcSummary <- list.files(path=baseDir, pattern="^summary.txt", full.names=TRUE, recursive=T) %>%  ldply(readSummary)
 
 #+ fig.height=20
-qcSummary %>% ggplot(aes(score, run, fill=tolower(flag))) +
+gg <- qcSummary %>% ggplot(aes(score, run, fill=tolower(flag))) +
     geom_tile() +
     rotXlab() +
-    scale_fill_manual(values = c(fail="darkred", pass="darkgreen", warn="orange"))
+    scale_fill_manual(values = c(fail="darkred", pass="darkgreen", warn="orange")) +
+    ggtitle("fastqc passcodes")
+
+#+ results='asis'
+gg %>%  ggsave2(width=15, height = round(nrow(readCounts)/4), limitsize=FALSE) %>% paste0("<img src='", ., "'><br>") %>% cat()
 
 
 #' # Base Quality Distribution Summary
@@ -98,7 +105,6 @@ baseQualities <- fastqDataFiles %>%  ldply(readBaseQualDist)
 #with(baseQualities, as.data.frame(table(run)))
 
 
-#+ fig.height=20
 baseQualities %>% ggplot(aes(reorder(Base, base_order), Mean, group=run, color=run)) + geom_line() + scale_y_continuous(limits=c(2, 40))