rliterman@0: // Screening and SNP Pipeline processing
rliterman@0: output_directory = file(params.output_directory)
rliterman@0: log_directory = file(params.log_directory)
rliterman@0: screen_log_dir = file(params.screen_log_dir)
rliterman@0: snp_log_dir = file(params.snp_log_dir)
rliterman@0: snp_directory = file(params.snp_directory)
rliterman@0: 
rliterman@0: if(params.tmp_dir == ""){
rliterman@0:     temp_dir = ""
rliterman@0: } else{
rliterman@0:     temp_dir = file(params.temp_dir)
rliterman@0: }
rliterman@0: ref_id_file = file(params.ref_id_file)
rliterman@0: 
rliterman@0: ref_mode = params.ref_mode
rliterman@0: 
rliterman@0: // Assess whether to rescue edge-filtered SNPs
rliterman@0: edge_rescue = "${params.rescue}" == "norescue" ? "norescue" : "rescue"
rliterman@0: 
rliterman@0: // Set paths for output files 
rliterman@0: all_snpdiffs_list = file("${log_directory}/All_SNPDiffs.txt")
rliterman@0: snp_dirs_list = file("${log_directory}/SNP_Dirs.txt")
rliterman@0: screening_results_file = file("${output_directory}/Screening_Results.tsv")
rliterman@0: isolate_data_file = file("${output_directory}/Isolate_Data.tsv")
rliterman@0: snpdiffs_summary_file = file("${output_directory}/Raw_MUMmer_Summary.tsv")
rliterman@0: 
rliterman@0: // Get QC thresholds
rliterman@0: min_cov = params.min_cov.toFloat()
rliterman@0: min_length = params.min_len.toInteger()
rliterman@0: min_iden = params.min_iden.toFloat()
rliterman@0: reference_edge = params.ref_edge.toInteger()
rliterman@0: query_edge = params.query_edge.toInteger()
rliterman@0: max_missing = params.max_missing.toFloat()
rliterman@0: n_ref = params.n_ref.toInteger()
rliterman@0: 
rliterman@0: workflow {
rliterman@0:     main:
rliterman@0:     // Run SNP pipeline
rliterman@0:     runSNPPipeline(query_data: all_snpdiffs, reference_data: ref_id_file)
rliterman@0: }
rliterman@0: 
rliterman@0: workflow runScreen {
rliterman@0:     
rliterman@0:     take:
rliterman@0:     all_snpdiffs
rliterman@0: 
rliterman@0:     main:
rliterman@0: 
rliterman@0:     all_snpdiffs
rliterman@0:     .unique{it -> it[2]} 
rliterman@0:     .collect() 
rliterman@0:     | screenSNPDiffs
rliterman@0: }
rliterman@0: 
rliterman@0: process screenSNPDiffs{
rliterman@0: 
rliterman@0:     input:
rliterman@0:     val(all_snpdiffs)
rliterman@0: 
rliterman@0:     script:
rliterman@0: 
rliterman@0:     screenDiffs = file("${projectDir}/bin/screenSNPDiffs.py")
rliterman@0:     """
rliterman@0:     $params.load_python_module
rliterman@0:     $params.load_bedtools_module
rliterman@0:     python $screenDiffs --snpdiffs_file "${all_snpdiffs_list}" --log_dir "${screen_log_dir}" --min_cov "${min_cov}" --min_len "${min_length}" --min_iden "${min_iden}" --ref_edge "${reference_edge}" --query_edge "${query_edge}" --density_windows "${params.dwin}" --max_snps "${params.wsnps}" --trim_name "${params.trim_name}" --output_file "${screening_results_file}" --ref_id "${ref_id_file}" --tmp_dir "${temp_dir}"
rliterman@0:     """
rliterman@0: } 
rliterman@0: 
rliterman@0: workflow runSNPPipeline{
rliterman@0:     take:
rliterman@0:     all_snpdiffs
rliterman@0:     reference_data
rliterman@0: 
rliterman@0:     main:
rliterman@0:     
rliterman@0:     query_snpdiffs = all_snpdiffs.map{tuple(it[0],it[2])}
rliterman@0:     ref_snpdiffs = all_snpdiffs.map{tuple(it[1],it[2])}
rliterman@0: 
rliterman@0:     stacked_snpdiffs = query_snpdiffs.concat(ref_snpdiffs)
rliterman@0:     .collect().flatten().collate(2)
rliterman@0: 
rliterman@0:     snp_dirs = stacked_snpdiffs
rliterman@0:     .combine(reference_data)
rliterman@0:     .filter{it -> it[0].toString() == it[2].toString()}
rliterman@0:     .map{it -> tuple(it[0],it[1])}
rliterman@0:     .groupTuple(by:0)
rliterman@0:     .map { ref, diff_files -> tuple( ref.toString(), diff_files.collect() ) }
rliterman@0:     | runSnpPipeline
rliterman@0: 
rliterman@0:     //snp_dirs.collect() | compileResults
rliterman@0: }
rliterman@0: 
rliterman@0: process compileResults{
rliterman@0: 
rliterman@0:     executor = 'local'
rliterman@0:     cpus = 1
rliterman@0:     maxForks = 1
rliterman@0:     
rliterman@0:     input:
rliterman@0:     val(snp_directories)
rliterman@0: 
rliterman@0:     script:
rliterman@0: 
rliterman@0:     compile_script = file("${projectDir}/bin/compileSNPResults.py")
rliterman@0:     snp_dirs_list.write(snp_directories.join("\n")+ "\n")
rliterman@0:     """
rliterman@0:     $params.load_python_module
rliterman@0:     python $compile_script --snp_dirs_file "${snp_dirs_list}" --output_directory "${snp_directory}" --isolate_data_file "${isolate_data_file}" --mummer_data_file "${snpdiffs_summary_file}"
rliterman@0:     """
rliterman@0: }
rliterman@0: 
rliterman@0: process runSnpPipeline{
rliterman@0: 
rliterman@0:     input:
rliterman@0:     tuple val(reference_id),val(diff_files)
rliterman@0: 
rliterman@0:     output:
rliterman@0:     stdout
rliterman@0: 
rliterman@0:     script:
rliterman@0: 
rliterman@0:     snp_script = file("${projectDir}/bin/runSNPPipeline.py")
rliterman@0: 
rliterman@0:     // Set + create output directory
rliterman@0:     snp_dir = file("${snp_directory}/${reference_id}")
rliterman@0:     snp_dir.mkdirs()
rliterman@0: 
rliterman@0:     // Write SNPDiffs list
rliterman@0:     out_snpdiffs = file("${snp_dir}/SNPDiffs.txt")
rliterman@0:     out_snpdiffs.write(diff_files.join("\n")+ "\n")
rliterman@0:     """
rliterman@0:     $params.load_python_module
rliterman@0:     $params.load_bedtools_module
rliterman@0:     python $snp_script --reference_id "${reference_id}" --output_directory "${snp_dir}" --snpdiffs_file "${out_snpdiffs}" --log_directory "${snp_log_dir}" --min_cov "${min_cov}" --min_len "${min_length}" --min_iden "${min_iden}" --ref_edge "${reference_edge}" --query_edge "${query_edge}" --density_windows "${params.dwin}" --max_snps "${params.wsnps}" --trim_name "${params.trim_name}" --max_missing "${max_missing}" --tmp_dir "${temp_dir}" --rescue "${edge_rescue}"
rliterman@0:     echo -n $snp_dir
rliterman@0:     """
rliterman@0: }