generate_config works with precomp buscos

Author: tkchafin

bin/generate_config.py

@@ -47,6 +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("--version", action="version", version="%(prog)s 2.0")
     return parser.parse_args(args)
 
@@ -121,20 +122,28 @@ def get_classification(taxon_info: TaxonInfo) -> typing.Dict[str, str]:
     return {r: ancestors[r] for r in RANKS if r in ancestors}
 
 
-def get_odb(taxon_info: TaxonInfo, lineage_tax_ids: str, requested_buscos: typing.Optional[str]) -> typing.List[str]:
+def get_odb(taxon_info: TaxonInfo, lineage_tax_ids: str, requested_buscos: typing.Optional[str], pre_computed_buscos: typing.List[str]) -> typing.List[str]:
     # Read the mapping between the BUSCO lineages and their taxon_id
     with open(lineage_tax_ids) as file_in:
         lineage_tax_ids_dict: typing.Dict[int, str] = {}
         for line in file_in:
             arr = line.split()
             lineage_tax_ids_dict[int(arr[0])] = arr[1] + "_odb10"
 
-    if requested_buscos:
+    valid_odbs = set(lineage_tax_ids_dict.values())
+
+    if pre_computed_buscos:
+        # Use pre-computed BUSCO lineages if available
+        odb_arr = pre_computed_buscos
+        for odb in odb_arr:
+            if odb not in valid_odbs:
+                print(f"Invalid pre-computed BUSCO lineage: {odb}", file=sys.stderr)
+                sys.exit(1)
+    elif requested_buscos:
         odb_arr = requested_buscos.split(",")
-        valid_odbs = set(lineage_tax_ids_dict.values())
         for odb in odb_arr:
             if odb not in valid_odbs:
-                print(f"Invalid BUSCO lineage: {odb}", file=sys.stderr)
+                print(f"Invalid requested BUSCO lineage: {odb}", file=sys.stderr)
                 sys.exit(1)
     else:
         # Do the intersection to find the ancestors that have a BUSCO lineage
@@ -327,7 +336,9 @@ def main(args=None):
 
     taxon_info = fetch_taxon_info(args.taxon_query)
     classification = get_classification(taxon_info)
-    odb_arr = get_odb(taxon_info, args.lineage_tax_ids, args.busco)
+
+    precomputed_busco = [os.path.basename(path).replace("run_", "") for path in args.busco_output]
+    odb_arr = get_odb(taxon_info, args.lineage_tax_ids, args.busco, precomputed_busco)
     taxon_id = adjust_taxon_id(args.nt, taxon_info)
 
     if sequence_report:

modules/local/generate_config.nf

@@ -15,6 +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)
 
     output:
     tuple val(meta), path("*.yaml")          , emit: yaml
@@ -32,6 +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}"
     """
     generate_config.py \\
         --fasta $fasta \\
@@ -45,6 +47,7 @@ process GENERATE_CONFIG {
         --blastx ${blastx} \\
         --blastn ${blastn} \\
         --taxdump ${taxdump} \\
+        $busco_output_param \\
         --output_prefix ${prefix}
 
     cat <<-END_VERSIONS > versions.yml

subworkflows/local/input_check.nf

@@ -132,8 +132,6 @@ workflow INPUT_CHECK {
         ch_parsed_busco = Channel.empty()
     }
 
-    // View the parsed channel
-    ch_parsed_busco.view()
 
     GENERATE_CONFIG (
         fasta,
@@ -145,6 +143,7 @@ workflow INPUT_CHECK {
         ch_databases.blastx,
         ch_databases.blastn,
         ch_databases.taxdump,
+        ch_parsed_busco.map { meta, path -> [meta, path] }.groupTuple()
     )
     ch_versions = ch_versions.mix(GENERATE_CONFIG.out.versions.first())