script

#!/bin/bash

set -e
set -u

#install usearch, vseach, fastQC
export usearch=path/to/usearch11.0.667_i86linux32

if [ x$usearch == x ] ; then
	echo Must set \$usearch >> /dev/stderr
	exit 1
fi

cd ~/PRJNA423040
rm -rf ./output
mkdir -p ./output
cd ./output

SECONDS=0

echo "STEP1: assembling reads..."

# we have single end reads
cat ../*.fastq >> merged.fastq

mkdir -p fastqc_check_before
echo "STEP2:running FAST QC on merged raw fastq files..."
fastqc merged.fastq -o ./fastqc_check_before

echo "STEP3: filtering the reads..."
# Strip primer (CCGTCAATTCMTTTRAGT, len 18,  strips 30 bases as stated)
$usearch -fastx_truncate merged.fastq -stripleft 30 -fastqout reads_stripped.fastq
$usearch -fastq_filter reads_stripped.fastq -fastq_maxee 1.0 -fastq_trunclen 350 -fastaout filtered.fa

echo "STEP4:dereplication: findind the unique sequences and abundances..."
vsearch --derep_fulllength filtered.fa -sizeout -relabel Uniq -output uniques.fa

echo "STEP5:picking OTUs and checking chimeras and discarding...it makes 97% OTUs"
$usearch -cluster_otus uniques.fa -minsize 2 -otus otus.fa -relabel Otu \
 -uparseout cluster_otus_detail.up -log cluster_otus.log

echo "STEP6:creating OTU tables..."
$usearch -otutab filtered.fa -otus otus.fa -otutabout otutab_raw.txt

echo "STEP7:assigning taxonomy to the OTU table..."
db_path="path/to/rdp_16s_v18.fa"
$usearch -sintax otus.fa -db "$db_path" -strand both \
  -tabbedout sintax.txt -sintax_cutoff 0.8

echo %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
echo ANALYSIS COMPLETE
echo %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

duration=$SECONDS
echo "Installing took $(($duration / 3600)) hours, \
 $(($duration / 60)) minutes and $(($duration % 60)) seconds."