phla

#!/bin/bash
# THREADED HLA typing off 10x assemblies
#inputs are a file that has the HLA region extracted from the megabubbles file
#(ie 'hla.bubbles.fasta'), hla/mhc reference region,  the HLA alleles query against the megabubbles, a name for your run, and
#the number of threads to us with pblat
## please use full paths!

#help output
if [ "$1" == "-h" ]; then
 echo "PLEASE USE ABSOLUTE PATHS!"
 echo "Usage: `basename $0` [1)hla.bubbles.fa 2)hla.genome.ref.fa 3)alleles-to-query.fa 4)name-of-run 5)number-of-threads]"
  exit 0
fi

#make sure there is an inputfile
if [ $# -eq 0 ]
  then
    echo "No arguments supplied"
    echo "Use -h for info"
    exit 1
fi

echo "------PERFORMING NUCMER AND MAKING MUMMER PLOT------"
# run nucmer and make the mummer plot to look at phasing across the HLA region
mkdir nucmer
cd nucmer

#nucmer -c 1000 -l 20 $1 $2
nucmer -l 40 -c 150 $1 $2
mummerplot -s large -l -t png -p $4  out.delta
gnuplot $4.gp
show-coords -c out.delta > out.coords
Rscript -e 'source("/mnt/home/stephen/Apps/bin/dotplot.R")'

cd ../

echo "------PERFORMING HLA TYPING------"
pblat -noHead -stepSize=2 $1 $3 -dots=100 -threads=$5 -out=psl $4.nuc.psl

perl /mnt/home/stephen/Apps/bin/pslScore.pl $4.nuc.psl > $4.nuc.scores.out

paste $4.nuc.psl $4.nuc.scores.out > $4.nuc.pls.scores.out

cut -f -17,26- $4.nuc.pls.scores.out | cat /mnt/home/stephen/Apps/bin/Headers/blat.short.header.tabs - > $4.nuc.pls.scores.final.out

echo "-----DONE-----"