feat: BUSCO auto lineage mode, tempdir, updated version (#526)

* Added tempdir to store all outputs that are copied as requested, options to specify dataset_dir, and new example

* Fixed linting

* Added full_table and missing_busco to output

* Fixed typo

* Added lineage checks

* Update bio/busco/environment.yaml

Co-authored-by: Johannes Köster <johannes.koester@uni-due.de>

bio/busco/environment.yaml

@@ -3,5 +3,4 @@ channels:
   - bioconda
   - nodefaults
 dependencies:
-  - busco==5.1.2
-
+  - busco =5.4

bio/busco/meta.yaml

@@ -1,9 +1,15 @@
 name: busco
 description: |
  Assess assembly and annotation completeness with BUSCO 
+url: https://busco.ezlab.org/
 authors:
   - Tessa Pierce
+  - Filipe G. Vieira
 input:
   - assembly fasta
 output:
   - annotation quality files
+notes: |
+  * The `mode` parameter sets the assessment mode (mandatory), and only allows for arguments: "genome", "transcriptome", and "proteins".
+  * The `lineage` parameter sets the lineage dataset ot use (optional). In `auto-lineage` mode, `output` or `dataset` folders need to be retrieved as a whole, since it is not possible to infer the output file names (they depend on the best lineage match).
+  * The `extra` param allows for additional program arguments.

bio/busco/test/Snakefile

@@ -1,16 +1,53 @@
 rule run_busco:
     input:
-        "transcripts.fasta"
+        "protein.fasta",
     output:
-        directory("txome_busco")
+        short_json="txome_busco/short_summary.json",
+        short_txt="txome_busco/short_summary.txt",
+        full_table="txome_busco/full_table.tsv",
+        miss_list="txome_busco/busco_missing.tsv",
+        dataset_dir=directory("resources/busco_downloads"),
     log:
-        "logs/quality/transcriptome_busco.log"
-    threads: 8
+        "logs/proteins_busco.log",
     params:
-        mode="transcriptome",
+        mode="proteins",
         lineage="stramenopiles_odb10",
-        downloads_path="resources/busco_downloads",
         # optional parameters
-        extra="" 
+        extra="",
+    threads: 8
+    wrapper:
+        "master/bio/busco"
+
+
+rule run_busco_euk:
+    input:
+        "protein.fasta",
+    output:
+        out_dir=directory("txome_busco/euk"),
+        dataset_dir=directory("resources/busco_downloads"),
+    log:
+        "logs/proteins_busco_euk.log",
+    params:
+        mode="proteins",
+        # optional parameters
+        extra="--auto-lineage-euk",
+    threads: 8
+    wrapper:
+        "master/bio/busco"
+
+
+rule run_busco_prok:
+    input:
+        "protein.fasta",
+    output:
+        out_dir=directory("txome_busco/prok"),
+        dataset_dir=directory("resources/busco_downloads"),
+    log:
+        "logs/proteins_busco_prok.log",
+    params:
+        mode="proteins",
+        # optional parameters
+        extra="--auto-lineage-prok",
+    threads: 8
     wrapper:
         "master/bio/busco"