added merging, deduplication

16S_workflow.sh

@@ -2,18 +2,18 @@
 raw_reads_dir=$1
 output_dir=$2
 
-#mkdir $output_dir
-#mkdir trimmed_reads
+mkdir $output_dir
+mkdir trimmed_reads
 #TRIM READS WITH ALIENTRIMMER
-#for R1 in $(ls $raw_reads_dir/*R1.fastq);
-#    do
-#        R2=`echo $R1 | sed 's/R1/R2/g'`
-#        oR1=`echo $R1 | sed 's/\.fastq/\.at\.fq/g'`
-#        oR1=`echo $oR1 | sed 's/fastq\///g'`
-#        oR2=`echo $R2 | sed 's/\.fastq/\.at\.fq/g'`
-#        oR2=`echo $oR2 | sed 's/fastq\///g'`
-#        java -jar soft/AlienTrimmer.jar -if $R1 -ir $R2 -q 20 -c databases/contaminants.fasta -of trimmed_reads/$oR1 -or trimmed_reads/$oR2
-#    done
+for R1 in $(ls $raw_reads_dir/*R1.fastq);
+    do
+        R2=`echo $R1 | sed 's/R1/R2/g'`
+        oR1=`echo $R1 | sed 's/\.fastq/\.at\.fq/g'`
+        oR1=`echo $oR1 | sed 's/fastq\///g'`
+        oR2=`echo $R2 | sed 's/\.fastq/\.at\.fq/g'`
+        oR2=`echo $oR2 | sed 's/fastq\///g'`
+        java -jar soft/AlienTrimmer.jar -if $R1 -ir $R2 -q 20 -c databases/contaminants.fasta -of trimmed_reads/$oR1 -or trimmed_reads/$oR2
+    done
 
 
 # MERGE READS AND OUTPUT FASTA
@@ -31,4 +31,15 @@ for R1 in $(ls trimmed_reads/*R1.at.fq);
 # MERGE ALL FASTAS TO 'amplicon.fasta'
 cat fasta/*.fasta > fasta/amplicon.fasta
 
+# REMOVE ' ' FROM EACH READ NAME
+sed -i 's/ //g' fasta/amplicon.fasta
+
+# CREATE OTU
+# DEDUPLICATION
+mkdir otu
+soft/vsearch --derep_fulllength fasta/amplicon.fasta --output otu/fulllength_otu
+soft/vsearch --derep_prefix fasta/amplicon.fasta --output otu/prefix_otu
 
+# REMOVING SINGLETON
+soft/vsearch --derep_fulllength fasta/amplicon.fasta --output otu/fulllength_min10_otu --minuniquesize 10
+soft/vsearch --derep_prefix fasta/amplicon.fasta --output otu/prefix_min10_otu --minuniquesize 10