annotate run_segalign_diagonal_partition @ 14:f9e91add973c draft

planemo upload for repository https://github.com/richard-burhans/galaxytools/tree/main/tools/segalign commit eb6ee047d39760f2596b6e343e2dc3898fafc5cd

#!/usr/bin/env bash

set -o errexit
set -o nounset
set -o pipefail

##
## parse arguments
##

SEGALIGN_ARGS=()

MAX_SEGMENT_SIZE=""
OUTPUT_FILENAME=""
LASTZ_COMMAND_FILE="lastz_commands.txt"
HELP=0

while [[ $# -gt 0 ]]; do
    case $1 in
        --help)
            HELP=1
            shift
            ;;
        --max_segments)
            MAX_SEGMENT_SIZE="$2"
            shift 2
            ;;
        --output)
            OUTPUT_FILENAME=$(readlink -f "$2")
            shift 2
            ;;
        --tool_directory)
            TOOL_DIRECTORY="$2"
            shift 2
            ;;
        *)
            SEGALIGN_ARGS+=("$1")
            shift
    esac
done

set -- "${SEGALIGN_ARGS[@]}"

##
## check arguments
##

if [[ $# == 0 || "$HELP" == "1" ]]; then
    segalign --help
    exit 0
fi

if [[ $# -lt 2 ]]; then
    echo "run_segalign_diagonal_partition: missing target and query sequences" 1>&2
    exit 1
fi

ref_path=$(readlink -f "$1")
test -e "$ref_path" || {
    echo "run_segalign_diagonal_partition: target file \"$ref_path\" does not exist" 1>&2
    exit 1
}
query_path=$(readlink -f "$2")
test -e "$query_path" || {
    echo "run_segalign_diagonal_partition: query file \"$query_path\" does not exist" 1>&2
    exit 1
}
shift 2

DATA_FOLDER="data"
mkdir -p "$DATA_FOLDER" || {
    echo "run_segalign_diagonal_partition: cannont create data directory \"$DATA_FOLDER\"" 1>&2
    exit 1
}

cd $DATA_FOLDER/..
echo ""
echo "Converting fasta files to 2bit format" 1>&2

##
## convert target and query to 2bit
##
faToTwoBit <(gzip -cdfq "$ref_path") "$DATA_FOLDER/ref.2bit" || {
    echo "run_segalign_diagonal_partition: cannot convert \"$ref_path\" to 2bit" 1>&2
    exit 1
}
faToTwoBit <(gzip -cdfq "$query_path") "$DATA_FOLDER/query.2bit" || {
    echo "run_segalign_diagonal_partition: cannont convert \"$ref_path\" to 2bit" 1>&2
    exit 1
}



time {
    while IFS= read -r line; do
        python "$TOOL_DIRECTORY/diagonal_partition.py" $MAX_SEGMENT_SIZE $line >> $LASTZ_COMMAND_FILE || {
            echo "run_segalign_diagonal_partition: failed: diagonal_partition.py $MAX_SEGMENT_SIZE $line" 1>&2
            exit 1
        }
        # segalign sort writes out the partitioned segment files to the working
        # directory and prints the modified lastz commands.
    done < <(stdbuf -oL segalign $ref_path $query_path "${DATA_FOLDER}/" "$@" ) # segalign begins running in this line, 
} 1>&2 # and every newline written to stdout, get assigned to $line which
       # gets sent to diagonal_partition for diagonal partitioning

exit 0