estimateGenomeSize

#! /usr/bin/env bash

if [[ -z $1 ]]; then
    echo "Perform the Kmer-frequency genome estimate method on a range of Kmer sizes"
    echo "Requires the software Jellyfish: https://github.com/gmarcais/Jellyfish"
    echo ""
    echo "[usage] estimateGenomeSize <# threads> <K start> <K end> <output prefix> <fastq.forward> <fastq.reverse>"
    echo "[example] estimateGenomeSize 10 17 22 tripletail TT_001.F.fq.gz TT_001.R.fq.gz"
    exit 1
fi

#$1 is the number of threads
#$2 is the range start
#$3 is the range end
#$4 is outfile prefix
#$5 is forward reads
#$6 is reverse reads

for i in $(seq $2 $3); do
    echo "Running jellyfish on k = $i"
    if (file $5 | grep -q compressed ) ; then
        jellyfish count -t $1 -C -m $i -s 7G -o $4_${i}mer <(zcat $5) <(zcat $6)
    else
        jellyfish count -t $1 -C -m $i -s 7G -o $4_${i}mer $5 $6
    fi
    echo "Jellyfish histo on k = $i"
    jellyfish histo -o $4_${i}mer.hist $4_${i}mer && rm $4_${i}mer
done