add

Trinity

@@ -214,6 +214,8 @@ my $MAX_CHRYSALIS_CLUSTER_SIZE = 25; # might need to set higher for highly polym
 
 my $JUST_NORMALIZE_READS_FLAG = 0;
 
+my $NO_SEQTK = 0;
+
 my $trinity_banner = qq^
      ______  ____   ____  ____   ____  ______  __ __
     |      ||    \\ |    ||    \\ |    ||      ||  |  |
@@ -542,6 +544,9 @@ my $advanced_usage =  <<_ADVANCEDUSAGE_;
 #  --stomp_snps                    : stomp snps out of kmers before inchworm assembly
 #
 #  --NO_SUPERTRANS                 : disable supertranscripts
+#
+#  --NO_SEQTK                      :disable seqtk for fq->fa conversions, instead use slower perl code
+#
 
 _ADVANCEDUSAGE_
 
@@ -759,6 +764,8 @@ my $STOMP_SNPS = 0;
     "FORCE" => \$FORCE_FLAG,
 
     "NO_SUPERTRANS" => \$NO_SUPERTRANS_FLAG,
+
+    "NO_SEQTK" => \$NO_SEQTK,
 
     );
 
@@ -2573,30 +2580,45 @@ sub prep_seqs {
 
             # make fasta
             foreach my $f (@initial_files){
-                my $cmd = "cat $f | seqtk-trinity seq -A -";
-                my $linecount_cmd = "cat $f | wc -l";
-                if ($f=~/\.gz$/){
-                    $cmd = "gunzip -c $f | seqtk-trinity seq -A -";
-                    $linecount_cmd = "gunzip -c $f | wc -l";
-                } elsif ($f=~/\.bz2$/){
-                    $cmd = "bunzip2 -dkc $f | seqtk-trinity seq -A -";
-                    $linecount_cmd = "bunzip2 -dkc $f | wc -l";
-                } elsif ($f =~ /\.xz/) {
-                    $cmd = "xz -dc ${f} | seqtk-trinity seq -A -";
-                    $linecount_cmd = "xz -dc ${f} | wc -l";
-                    ## I would like to suggest that these if statements are not necessary if one just does
-                    ## qx"less ${f} |" because less has smart input filters in place and will automagically
-                    ## handle all the likely compression formats.
-                }
 
-                if ($SS_lib_type && $SS_lib_type eq "R") {
-                    $cmd =~ s/trinity seq /trinity seq -r /;
+                if ($NO_SEQTK) {
+                    my $perlcmd = "$UTILDIR/support_scripts/fastQ_to_fastA.pl -I $f ";
+
+                    if ($SS_lib_type && $SS_lib_type eq "R") {
+                        $perlcmd .= " --rev ";
+
+                    }
+                    $perlcmd .= " >> $file_prefix.fa 2> $f.readcount ";
+
+                    &process_cmd($perlcmd);
                 }
+                else {
+                    ## using seqtk (trinity-mod'd version)
 
-                $cmd .= " >> $file_prefix.fa";
-
-                &process_cmd($cmd);
-
+                    my $cmd = "cat $f | seqtk-trinity seq -A -";
+                    my $linecount_cmd = "cat $f | wc -l";
+                    if ($f=~/\.gz$/){
+                        $cmd = "gunzip -c $f | seqtk-trinity seq -A -";
+                        $linecount_cmd = "gunzip -c $f | wc -l";
+                    } elsif ($f=~/\.bz2$/){
+                        $cmd = "bunzip2 -dkc $f | seqtk-trinity seq -A -";
+                        $linecount_cmd = "bunzip2 -dkc $f | wc -l";
+                    } elsif ($f =~ /\.xz/) {
+                        $cmd = "xz -dc ${f} | seqtk-trinity seq -A -";
+                        $linecount_cmd = "xz -dc ${f} | wc -l";
+                        ## I would like to suggest that these if statements are not necessary if one just does
+                        ## qx"less ${f} |" because less has smart input filters in place and will automagically
+                        ## handle all the likely compression formats.
+                    }
+
+                    if ($SS_lib_type && $SS_lib_type eq "R") {
+                        $cmd =~ s/trinity seq /trinity seq -r /;
+                    }
+
+                    $cmd .= " >> $file_prefix.fa";
+
+                    &process_cmd($cmd);
+                }
             }
 
         }
@@ -3225,6 +3247,10 @@ sub normalize {
     if ($NO_CLEANUP) {
         $cmd .= " --no_cleanup ";
     }
+    if ($NO_SEQTK) {
+        $cmd .= " --NO_SEQTK ";
+    }
+
 
     #@read_files = &add_fifo_for_gzip(@read_files);
 

sample_data/test_Trinity_Assembly/misc_run_tests/__runMe.noSeqtk.sh

@@ -0,0 +1,19 @@
+#!/bin/bash -ve
+
+#######################################################
+##  Run Trinity to Generate Transcriptome Assemblies ##
+#######################################################
+
+if [ -z ${TRINITY_HOME} ]; then
+    echo "Must set env var TRINITY_HOME"
+    exit 1
+fi
+
+
+${TRINITY_HOME}/Trinity --seqType fq --max_memory 2G \
+              --left reads.left.fq.gz \
+              --right reads.right.fq.gz \
+              --SS_lib_type RF \
+              --CPU 4 \
+              --output trinity_out_dir_noseqtk --NO_SEQTK
+

util/insilico_read_normalization.pl

@@ -52,6 +52,8 @@
 my $PARALLEL_STATS = 0;
 my $JELLY_S;
 
+my $NO_SEQTK = 0;
+
 my $usage = <<_EOUSAGE_;
 
 
@@ -164,6 +166,8 @@
     'jelly_s=i' => \$JELLY_S,
 
     'tmp_dir_name=s' => \$TMP_DIR_NAME, 
+
+    "NO_SEQTK" => \$NO_SEQTK,
 
 );
 
@@ -641,13 +645,28 @@ sub prep_seqs {
 
     if ($seqType eq "fq") {
         # make fasta
-        my $cmd = "seqtk-trinity seq -A";
-        if ($SS_lib_type && $SS_lib_type eq "R") {
-            $cmd  =~ s/trinity seq /trinity seq -r /;
+
+        if ($NO_SEQTK) {
+            my $perlcmd = "$UTILDIR/fastQ_to_fastA.pl -I $initial_file ";
+
+            if ($SS_lib_type && $SS_lib_type eq "R") {
+                $perlcmd .= " --rev ";
+
+            }
+            $perlcmd .= " >> $file_prefix.fa 2> $file_prefix.readcount ";
+
+            &process_cmd($perlcmd);
+        }
+        else {
+            # using seqtk
+            my $cmd = "seqtk-trinity seq -A";
+            if ($SS_lib_type && $SS_lib_type eq "R") {
+                $cmd  =~ s/trinity seq /trinity seq -r /;
+            }
+            $cmd .= " $initial_file >> $file_prefix.fa";
+
+            &process_cmd($cmd);
         }
-        $cmd .= " $initial_file >> $file_prefix.fa";
-
-        &process_cmd($cmd);
     }
     elsif ($seqType eq "fa") {
         if ($SS_lib_type && $SS_lib_type eq "R") {