diff sum_fastqc.pl @ 5:a2c2dc7bc724 draft

Uploaded

--- a/sum_fastqc.pl	Tue Oct 16 21:59:23 2018 -0400
+++ b/sum_fastqc.pl	Tue Oct 16 21:59:38 2018 -0400
@@ -11,37 +11,46 @@
 ##
 #################################################### 
 
-$infile=@ARGV[0];
+my($inname)=shift(@ARGV);
+my($qscore)=shift(@ARGV);
+
+print "Input\tFile\tFastQC\tPass-Fail\tReads\tPoor_Reads\tGC\tQ".$qscore."\n";
+
+foreach (@ARGV) {
+  print_stats($_);
+}
 
-# First 10 lines of raw FASTQC contain basic overview
-@sumlines=`head -n 10 $infile`;
-chomp(@sumlines);
+sub print_stats {
+  $infile = shift;
+  # First 10 lines of raw FASTQC contain basic overview
+  @sumlines=`head -n 10 $infile`;
+  chomp(@sumlines);
 
-# Sequence level Q scores are buried in the middle of the file
-@qlines=`awk '/#Quality\tCount/,/>>END_MODULE/' $infile | head -n -1 | tail -n +2`;
-chomp(@qlines);
+  # Sequence level Q scores are buried in the middle of the file
+  @qlines=`awk '/#Quality\tCount/,/>>END_MODULE/' $infile | head -n -1 | tail -n +2`;
+  chomp(@qlines);
 
-@fastqc = split(/[\n\t]/,shift(@sumlines));
-@pass = split(/\t/,shift(@sumlines));
-shift(@sumlines);
-@fn = split(/\t/,shift(@sumlines));
-shift(@sumlines);
-shift(@sumlines);
-@nreads = split(/\t/,shift(@sumlines));
-@npoor = split(/\t/,shift(@sumlines));
-shift(@sumlines);
-@gc = split(/\t/,shift(@sumlines));
+  @fastqc = split(/[\n\t]/,shift(@sumlines));
+  @pass = split(/\t/,shift(@sumlines));
+  shift(@sumlines);
+  @fn = split(/\t/,shift(@sumlines));
+  shift(@sumlines);
+  shift(@sumlines);
+  @nreads = split(/\t/,shift(@sumlines));
+  @npoor = split(/\t/,shift(@sumlines));
+  shift(@sumlines);
+  @gc = split(/\t/,shift(@sumlines));
 
-print "File\tFastQC\tPass-Fail\tReads\tPoor_Reads\tGC\tQ30\n";
-
-print $fn[1]."\t";
-print $fastqc[1]."\t";
-print $pass[1]."\t";
-print $nreads[1]."\t";
-print $npoor[1]."\t";
-print $gc[1]."\t";
-print qcal($nreads[1],30,\@qlines);
-print "\n";
+  print $inname."\t";
+  print $fn[1]."\t";
+  print $fastqc[1]."\t";
+  print $pass[1]."\t";
+  print $nreads[1]."\t";
+  print $npoor[1]."\t";
+  print $gc[1]."\t";
+  print qcal($nreads[1],$qscore,\@qlines);
+  print "\n";
+}
 
 # Sum reads w/ Q scores > cutoff and divide by number of reads
 sub qcal {
@@ -51,9 +60,9 @@
    $sum = 0;
   
    foreach $item (@qarray) {
-      ($qval,$qscore)=split(/\t/,$item);
+      my($qval,$q)=split(/\t/,$item);
       if($qval>=$cutoff) {
-        $sum += $qscore;
+        $sum += $q;
       }
    }
    $qmean = sprintf("%.2f", 100 * $sum / $nreads);