Find likely adapter sequences in a set of paired end FASTQs

find_adapters.sh

#!/bin/bash

# Given a directory of fastq files, use bbmerge to find adapters sequences for every sample,
# align and find the consensus. 
# Final consensus is in  r1_adapter.consensus.fa and r2_adapter.consensus.fa - trim by hand
# if the automatic trimming up the the first 'n' isn't sensible

# Requires: bbmerge.sh (bbmap), muscle, emboss (cons)


# MODIFY to match your R1/R2 suffix ()
r1_suffix="_L2_1.fq.gz"
r1_tail="_1.fq.gz"
r2_tail="_2.fq.gz"

module load bbmap || true

for r1 in ${1}/*${r1_tail}; do 
  r2="${r1/${r1_tail}/${r2_tail}}"
  bbmerge.sh in=${r1} in2=${r2} outa=adapters_bbmerge.$(basename ${r1} ${r1_suffix}).fa
done


echo -n "" >r1_adapters.fa
echo -n "" >r2_adapters.fa

for f in adapters_bbmerge.*.fa; do
  head -n 2 $f >>r1_adapters.fa
  tail -n 2 $f >>r2_adapters.fa
done

module load muscle || true

muscle -align r1_adapters.fa -output r1_adapters.muscle.fa

muscle -align r2_adapters.fa -output r2_adapters.muscle.fa

# from the EMBOSSS suite
module load emboss || true

# Find the consensus sequence
cons -sequence r1_adapters.muscle.fa -outseq r1_adapter.consensus.fa -name R1_adapter
cat r2_adapter.consensus.fa
# Turn multi-line FASTA into single line, output everything up to the first 'n'
echo "Trimmed:"
awk '/^>/ {if(seq) print seq; print; seq=""; next} {seq = seq $0} END {print seq}' r1_adapter.consensus.fa | sed 's/n\{1,\}.*//'

# Find the consensus sequence
cons -sequence r2_adapters.muscle.fa -outseq r2_adapter.consensus.fa -name R2_adapter
cat r2_adapter.consensus.fa
# Turn multi-line FASTA into single line, output everything up to the first 'n'
echo "Trimmed:"
awk '/^>/ {if(seq) print seq; print; seq=""; next} {seq = seq $0} END {print seq}' r2_adapter.consensus.fa | sed 's/n\{1,\}.*//'