fixed generate_config with or without precomp busco

Author: tkchafin

bin/generate_config.py

@@ -47,7 +47,7 @@ def parse_args(args=None):
     parser.add_argument("--blastx", help="Path to the blastx database", required=True)
     parser.add_argument("--blastn", help="Path to the blastn database", required=True)
     parser.add_argument("--taxdump", help="Path to the taxonomy database", required=True)
-    parser.add_argument("--busco_output", action="append", help="Path to BUSCO output directory", required=True)
+    parser.add_argument("--busco_output", action="append", help="Path to BUSCO output directory", required=False)
     parser.add_argument("--version", action="version", version="%(prog)s 2.0")
     return parser.parse_args(args)
 
@@ -337,7 +337,7 @@ def main(args=None):
     taxon_info = fetch_taxon_info(args.taxon_query)
     classification = get_classification(taxon_info)
 
-    precomputed_busco = [os.path.basename(path).replace("run_", "") for path in args.busco_output]
+    precomputed_busco = [os.path.basename(path).replace("run_", "") for path in (args.busco_output or [])]
     odb_arr = get_odb(taxon_info, args.lineage_tax_ids, args.busco, precomputed_busco)
     taxon_id = adjust_taxon_id(args.nt, taxon_info)
 

modules/local/generate_config.nf

@@ -15,7 +15,7 @@ process GENERATE_CONFIG {
     tuple val(meta3), path(blastx, stageAs: 'blastx/*')
     tuple val(meta4), path(blastn, stageAs: 'blastn/*')
     tuple val(meta5), path(taxdump)
-    tuple val(busco_meta), path(busco_output)
+    val (busco_outputs)
 
     output:
     tuple val(meta), path("*.yaml")          , emit: yaml
@@ -33,7 +33,7 @@ process GENERATE_CONFIG {
     def busco_param = busco_lin ? "--busco '${busco_lin}'" : ""
     def accession_params = params.accession ? "--accession ${params.accession}" : ""
     def input_reads = reads.collect{"--read_id ${it[0].id} --read_type ${it[0].datatype} --read_layout ${it[0].layout} --read_path ${it[1]}"}.join(' ')
-    def busco_output_param = "--busco_output ${busco_output}"
+    def busco_output_param = busco_outputs.collect { meta, path -> "--busco_output ${path}" }.join(' ')
     """
     generate_config.py \\
         --fasta $fasta \\

subworkflows/local/input_check.nf

@@ -143,7 +143,7 @@ workflow INPUT_CHECK {
         ch_databases.blastx,
         ch_databases.blastn,
         ch_databases.taxdump,
-        ch_parsed_busco.map { meta, path -> [meta, path] }.groupTuple()
+        ch_parsed_busco.toList()
     )
     ch_versions = ch_versions.mix(GENERATE_CONFIG.out.versions.first())
 
@@ -178,6 +178,16 @@ workflow INPUT_CHECK {
     ch_parsed_csv.busco_lineage
     | collect
     | set { ch_busco_lineages }
+    ch_busco_lineages.view()
+    ch_parsed_busco.view()
+
+    // Remove any invalid lineages from busco_outputs
+    ch_busco_lineages_list = ch_busco_lineages.flatten()
+    ch_parsed_busco_filtered = ch_parsed_busco
+        .filter { meta, path ->
+            ch_busco_lineages.contains(meta.lineage)
+        }
+    ch_parsed_busco_filtered = ch_parsed_busco_filtered.ifEmpty { Channel.empty() }
 
     emit:
     reads                                                        // channel: [ val(meta), path(datafile) ]
@@ -189,7 +199,7 @@ workflow INPUT_CHECK {
     blastn = ch_databases.blastn                                 // channel: [ val(meta), path(blastn_db) ]
     blastp = ch_databases.blastp                                 // channel: [ val(meta), path(blastp_db) ]
     blastx = ch_databases.blastx                                 // channel: [ val(meta), path(blastx_db) ]
-    busco_output = Channel.empty()                     // channel: [ val(meta), path(busco_output) ]
+    busco_output = ch_parsed_busco_filtered                      // channel: [ val(meta), path(busco_output) ]
     busco_db = ch_databases.busco.map { _, db_path -> db_path }  // channel: [ path(busco_db) ]
     taxdump = ch_databases.taxdump.map { _, db_path -> db_path } // channel: [ path(taxdump) ]
     versions = ch_versions                                       // channel: [ versions.yml ]