blca_multiscreen.sh

#!/bin/bash
numeric='^[0-9]+$'

source conda.sh # CHANGE to path of conda.sh
conda activate BLCA_env # CHANGE if environment name is different

### validate arguments ----

correct_usage="Usage: $0 [path to base file] [number of split files]"

if [[ ( $* == "--help") ||  $* == "-h" || $# -le 1 ]]
then 
	echo "$correct_usage"
	exit 0
fi

base_name=$(basename ${1})

ext=$(echo "$base_name" | awk -F. '{print ($NF)}') # use '.' as delimiter; extract lowercase file extension
pref=$(echo "$base_name" | awk -F. '{print ($1)}')

if ! [[ ".$ext" = ".fasta" ]] && ! [[ ".$ext" = ".fa" ]]
then
  printf "Error: supply the path to a file with one of the following extensions: .fasta, .fa. \n%s$ext was provided.\n"
  echo "$correct_usage"
  exit 0
elif ! [[ $2 =~ $numeric ]]
then
  echo "Error: supply numeric as second argument. $2 was provided."
  echo "$correct_usage"
  exit 0
fi

### ----

base_filepath=$1 # full (absolute) path to provided file

base_filedir="${base_filepath%/*}/" # directory that houses the provided file

declare -i num_files="$2" # convert to numeric

printf "Path to input file: $base_filepath\n"

# main folder for output
if ! mkdir "${base_filedir}BLCA_output"; then
  printf "The folder <<BLCA_output>> already exists. Please rename or move this folder to avoid conflicts.\n"
  exit 1
fi

mkdir "${base_filedir}BLCA_output/files" # folder to hold BLCA files
mkdir "${base_filedir}BLCA_output/logs" # folder to hold screen logfiles

pyfasta split -n "$num_files" "$base_filepath"

file_names=()
for ((i=0; i<num_files; i++)); do
  if [ "$i" -ge 10 ]; then
      file_names[i]=$(printf "%sBLCA_output/files/%s.%d.%s" "$base_filedir" "$pref" "$i" "$ext")
      location=$(printf "%s%s.%d.%s" "$base_filedir" "$pref" "$i" "$ext")
      mv $location "$base_filedir/BLCA_output/files"
  else
      file_names[i]=$(printf "%sBLCA_output/files/%s.0%d.%s" "$base_filedir" "$pref" "$i" "$ext")
      location=$(printf "%s%s.0%d.%s" "$base_filedir" "$pref" "$i" "$ext")
      mv $location "$base_filedir/BLCA_output/files"
  fi
done

printf "\nSplit files are in ${base_filedir}BLCA_output/files\n"

# open screen sessions (sessions are created and detached)
# CHANGE location of screen_command.sh if necessary
for ((i=0; i<num_files; i++)); do
  screen -d -m -S "BLCA_task_$i" -L -Logfile "$base_filedir/BLCA_output/logs/out_BLCA_task_$i.txt" screen_command.sh "${file_names[i]}"
done

done=0
printf "\nWaiting for all tasks to finish...\n"
while [[ $done -ne 1 ]]; do
  if !(screen -list | grep -q "BLCA_task_"); then
    done=1
  fi
  sleep 1
done

cd $base_filedir

mv *.$ext."gdx" "$base_filedir/BLCA_output/files"
mv *.$ext."flat" "$base_filedir/BLCA_output/files"

cd BLCA_output

cd files

# concatenate individual task files into 
# combined BLCA results for the entire input file
cat $pref.*.$ext."blastn" > ../combined.fa.blastn

cat $pref.*.$ext."blca.out" > ../combined.fa.blca.out

cd ..

# output info to general log

printf "file processed: $pref\n\n" >> BLCA_log.txt 

printf "number of sequences processed / BLCA output file:\n" >> BLCA_log.txt
wc -l combined.fa.blca.out >> BLCA_log.txt

cd files

printf "\nnumber of files generated by BLCA:\n" >> ../BLCA_log.txt
ls -1 | wc -l >> ../BLCA_log.txt

printf "\nTasks complete. All files are located in ${base_filedir}BLCA_output/\n"