pipeline.xml

<?xml version="1.0" encoding="UTF-8"?>

<pipeline xmlns="http://www.sing-group.org/compi/pipeline-1.0"
	xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
	<version>1.1.2</version>

	<params>
		<param name="working_dir" shortName="w" global="true" defaultValue="/working_dir">Path of the working directory.</param>
		<param name="input_dir" shortName="id" global="true" defaultValue="/input">Path of the input directory.</param>
		<param name="output_dir" shortName="od" global="true" defaultValue="/output">Path of the output directory.</param>
		<param name="blast_type" shortName="bt">BLAST type to use when finding orthologs.</param>
		<param name="reference_file" shortName="rf">Name of the reference genome file. This file must exist in the input_dir.</param>
	</params>

	<tasks>
		<task id="initialization" params="reference_file">
			if [ ! -f ${input_dir}/${reference_file} ]
			then
				echo "Error: the specified reference file (${reference_file}) can't be found in the input directory."
				exit -1
			fi
			mkdir -p ${working_dir}/temp -p ${working_dir}/files -p ${working_dir}/databases -p ${working_dir}/locks
		</task>
		<foreach id="prepare-files-and-dbs" after="initialization"
			of="command" in="ls ${input_dir}/*" as="file" params="blast_type">
				if [ -s ${file} ]
				then
					FILENAME=$(basename -- "${file}")

					sed '/^>/ s/$/#######/g' ${file} > ${working_dir}/temp/${FILENAME}.temp1
					sed -z 's/\r//g' ${working_dir}/temp/${FILENAME}.temp1 > ${working_dir}/temp/${FILENAME}.temp2
					sed -z 's/\n//g' ${working_dir}/temp/${FILENAME}.temp2 > ${working_dir}/temp/${FILENAME}.temp3
					sed -i 's/>/\n>/g' ${working_dir}/temp/${FILENAME}.temp3
					sed -i 's/#######/\n/g' ${working_dir}/temp/${FILENAME}.temp3
					sed -i '1d' ${working_dir}/temp/${FILENAME}.temp3

					mv  ${working_dir}/temp/${FILENAME}.temp3 ${working_dir}/files/${FILENAME}
					rm ${working_dir}/temp/${FILENAME}.temp*

					makeblastdb -in ${working_dir}/files/${FILENAME} -dbtype nucl -parse_seqids -out ${working_dir}/databases/${FILENAME}
				fi
		</foreach>
		<task id="split-reference" after="prepare-files-and-dbs" params="reference_file">
            rm -rf ${working_dir}/reference_split
			seqkit split --by-size 1 ${working_dir}/files/${reference_file} --out-dir ${working_dir}/reference_split
		</task>
		<task id="create-combinations-file" after="split-reference">
			rm -f ${working_dir}/temp/combinations
			for ref_file in $(ls ${working_dir}/reference_split); do
				for input_file in $(ls ${working_dir}/files); do
					echo $ref_file $input_file >> ${working_dir}/temp/combinations
					touch ${working_dir}/locks/${ref_file}
				done 
			done
		</task>
		<foreach id="run-blast" after="create-combinations-file"
			of="command" in="cat ${working_dir}/temp/combinations" as="combination"
			params="blast_type reference_file">
			<![CDATA[
			reference=$(echo ${combination} | cut -f1 -d' ')
			file=$(echo ${combination} | cut -f2 -d' ')
			combination="${reference}_${file}"
			
			${blast_type} -query ${working_dir}/reference_split/${reference} -db ${working_dir}/databases/${file} -evalue 0.05 -max_target_seqs 1 -outfmt 6 -out ${working_dir}/temp/${combination}.output.1
			
			seqref=$(cut -f1 ${working_dir}/temp/${combination}.output.1 | head -1)
			seqID1=$(cut -f2 ${working_dir}/temp/${combination}.output.1 | head -1)

			if [ -z "$seqID1" ]
			then
				echo "No Blast hit"
			else
				echo "Got a Blast hit"
				grep -w -A1 $seqID1 ${working_dir}/files/${file} > ${working_dir}/temp/${combination}.seq

				${blast_type} -query ${working_dir}/temp/${combination}.seq -db ${working_dir}/databases/${reference_file} -evalue 0.05 -max_target_seqs 1 -outfmt 6 -out ${working_dir}/temp/${combination}.output.2

				seqID2=$(cut -f2 ${working_dir}/temp/${combination}.output.2 | head -1)
				if [ "$seqref" == "$seqID2" ]
				then
					( flock 99
						grep -w -A1 $seqID1 ${working_dir}/files/${file} >> ${output_dir}/$seqref.orthologs
					) 99<"${working_dir}/locks/${reference}"
				fi
			fi

			rm ${working_dir}/temp/${combination}.*
			]]>
		</foreach>
	</tasks>

	<metadata>
		<task-description id="initialization">Initializes the working directory, creating the neccessary intermediate directories for the other pipeline tasks.</task-description>
		<task-description id="prepare-files-and-dbs">Copies the input files to the working directory in order to prepare them and create the BLAST databases.</task-description>
		<task-description id="split-reference">Splits the reference file to create one file for each sequence contained in it.</task-description>
		<task-description id="create-combinations-file">Creates the combinations file containing all possible pairs of reference sequences and input files.</task-description>
		<task-description id="run-blast">For each pair of reference sequence and input file, runs the two-way BLAST method for finding orthologs, writing the outputs in the output directory.</task-description>
	</metadata>
</pipeline>