run_all

#!/usr/bin/env bash
#
#
#
#  Usage:    run_all [DIRECTORY]
#
#
#  Notes:   [DIRECTORY] must contained all of the samples to be analyized (see Description below)
#           [DIRECTORY] must be the name of the directory, without the "/"
#                       (i.e. "$ run_all all_samples" works, "$ run_all all_samples/" does not)
#
#  Description:
#     This script will take all of the pairs of samples in a file, separate them into folders,
#     and then run the assembly pipeline on the samples.
#  
#     So:
#  
#     $ tree all_samples
#
#     all_samples
#     |1A_CP05062_cat_R001.fastq.gz
#     |1A_CP05062_cat_R002.fastq.gz
#     |2_CP05062_cat_R001.fastq.gz
#     |2_CP05062_cat_R002.fastq.gz
#     (etc.)
#  
#     Turns into:
#     $ tree all_samples
#
#     all_samples
#     |--1A_CP05062
#       |--raw_data
#          |1A_CP05062_cat_R001.fastq.gz
#          |1A_CP05062_cat_R001.fastq.gz
#       |--reports
#       |--assembly
#       |--trimmed_data
#     |--2_CP05062
#       |--raw_data
#       (etc)
#
#
#     Then each folder and pair of reads is analyzed based on our protocol



if [[ $CONDA_DEFAULT_ENV = "" ]] then
  echo "You do not have conda activated! This is required for prokka. exiting..."
  exit 1
fi



# list each individual sample
for x in $1/*R1*;

do
  FILE=$(basename $x);      #name of the .fastq.gz file (e.g. 1A_CP05062_cat_R001.fastq.gz)
  SAMPLE=${FILE%_cat*}      #(e.g. 1A_CP05062)
  PREFIX=${SAMPLE%_*}       #(e.g. 1A)

  echo "making directory for $PREFIX...";
  DIR=$1/$SAMPLE
  mkdir -p $DIR/raw_data $DIR/trimmed_data $DIR/assembly $DIR/reports;
  mv $1/$PREFIX*.gz $DIR/raw_data;

  echo "running fastqc for $PREFIX...";
  fastqc -o $DIR/reports/ $DIR/raw_data/*R1* $DIR/raw_data/*R2*;

  echo "running trimmomatic for $PREFIX...";
  trimmomatic PE $DIR/raw_data/*R1* $DIR/raw_data/*R2* \
    $DIR/trimmed_data/"$PREFIX"_forward_paired.fq.gz $DIR/trimmed_data/"$PREFIX"_forward_unpaired.fq.gz \
    $DIR/trimmed_data/"$PREFIX"_reverse_paired.fq.gz $DIR/trimmed_data/"$PREFIX"_reverse_unpaired.fq.gz \
    ILLUMINACLIP:TruSeq3-PE.fa:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:36;

  echo "running spades for $PREFIX...";
  spades -1 $DIR/trimmed_data/*forward_paired* -2 $DIR/trimmed_data/*reverse_paired* -o $DIR/assembly/;

  echo "running QUAST for $PREFIX...";
  quast.py -o $DIR/reports/quast_results/ $DIR/assembly/contigs.fasta;

  echo "assembling for $PREFIX using Prokka...";
  prokka --outdir $DIR/annotation/ --prefix "$PREFIX"_genome $DIR/assembly/contigs.fasta;

  echo "done with $PREFIX...";

  done

echo "is... is it done? did it work?";