tblastn-hap2-script.pbs

#!/bin/bash
#PBS -q condo05
#PBS -l select=1:ncpus=24:mem=200GB
#PBS -l walltime=168:00:00
#PBS -W group_list=x-ccast-prj-hulke
#PBS -N tblastn_hap2
#PBS -j oe

# Set the working directory
cd /mmfs1/projects/brent.hulke/MutagenesisSmart/HudsonAlpha/ChrLabels17Scaffs3Assemblies

# Define the tblastn and makeblastdb executables
TBLASTN="/mmfs1/projects/brent.hulke/Software/ncbi-blast-2.16.0+/bin/tblastn"
MAKEBLASTDB="/mmfs1/projects/brent.hulke/Software/ncbi-blast-2.16.0+/bin/makeblastdb"

# Define genome assembly
GENOME="Hap2_TrioChromosomes_10Mb_17seqs_ChromosomeLabels_sorted.fa"

# Define protein files
PROTEINS=(
    "F3H.fa"
    "myb.fa"
    "FLS.fa"
    "CHI.fa"
    "DFR.fa"
    "ANS.fa"
    "CHS.fa"
)

# Create output directory
mkdir -p tblastn_results/hap2

# Create BLAST database for the genome
$MAKEBLASTDB -in "$GENOME" -dbtype nucl -out "${GENOME%.*}_db"

# Loop through each protein file
for protein in "${PROTEINS[@]}"; do
    output_file="tblastn_results/hap2/${protein%.*}_vs_${GENOME%.*}.out"
    
    # Run tblastn
    $TBLASTN -query "$protein" -db "${GENOME%.*}_db" -out "$output_file" \
             -num_threads 24 \
             -outfmt 6
    
    echo "Completed tblastn search: $protein vs $GENOME"
done

echo "All tblastn searches completed for Hap2 assembly."