fix: Expose gene annotation params

config/config.yaml

@@ -8,6 +8,9 @@ reference:
   build: GRCh38
   # Ensembl release
   release: 107
+  # flavor: ""  # optional, e.g. chr_patch_hapl_scaff, see Ensembl FTP: https://ftp.ensembl.org/pub/release-107/gtf/homo_sapiens/
+  # branch: ""  # optional
+
   # for available downloads, please browse either of these views:
   # * http://repeatmasker.org/genomicDatasets/RMGenomicDatasets.html
   # * http://repeatmasker.org/genomicDatasets/RMGenomicDatasetsAlt.html

workflow/rules/annotate.smk

@@ -26,9 +26,11 @@ rule cyrcular_annotate_graph:
         graph="results/calling/graphs/{group}.graph",
         gene_annotation="resources/gene_annotation.gff3.gz",
         regulatory_annotation="resources/regulatory_annotation.gff3.gz",
-        repeat_annotation=lambda wc: "resources/repeat_masker.fa.out.gz"
-        if config["reference"].get("repeat_masker_download_link", "")
-        else "",
+        repeat_annotation=lambda wc: (
+            "resources/repeat_masker.fa.out.gz"
+            if config["reference"].get("repeat_masker_download_link", "")
+            else ""
+        ),
     output:
         annotated="results/calling/graphs/{group}.annotated.graph",
     threads: 1
@@ -39,9 +41,11 @@ rule cyrcular_annotate_graph:
     conda:
         "../envs/cyrcular.yaml"
     params:
-        repeat_annotation=lambda wc, input: f"  --repeat-annotation {input.repeat_annotation} "
-        if config["reference"].get("repeat_masker_download_link", "")
-        else "",
+        repeat_annotation=lambda wc, input: (
+            f"  --repeat-annotation {input.repeat_annotation} "
+            if config["reference"].get("repeat_masker_download_link", "")
+            else ""
+        ),
     shell:
         "cyrcular graph annotate "
         "  --reference {input.reference} "

workflow/rules/map.smk

@@ -11,7 +11,7 @@ rule minimap2_bam:
     params:
         extra=get_minimap2_mapping_params,  # optional
         sorting="coordinate",  # optional: Enable sorting. Possible values: 'none', 'queryname' or 'coordinate'
-        sort_extra=lambda wc, threads: f"-@ {min(threads, 4)}",  # optional: extra arguments for samtools/picard
+        sort_extra=lambda wc, threads: f"-@ {min(threads , 4)}",  # optional: extra arguments for samtools/picard
     threads: workflow.cores // 2
     wrapper:
         "v1.25.0/bio/minimap2/aligner"
@@ -27,9 +27,11 @@ rule merge_fastqs:
     wildcard_constraints:
         read="single|R1|R2",
     params:
-        cmd=lambda wc: "pigz -dc"
-        if (any(map(lambda f: f.endswith(".gz"), get_fastqs(wc))))
-        else "cat",
+        cmd=lambda wc: (
+            "pigz -dc"
+            if (any(map(lambda f: f.endswith(".gz"), get_fastqs(wc))))
+            else "cat"
+        ),
     conda:
         "../envs/pigz.yaml"
     shell:

workflow/rules/ref.smk

@@ -45,8 +45,6 @@ rule minimap2_index:
         ),
     log:
         "logs/minimap2_index/genome.log",
-    benchmark:
-        "benchmarks/minimap2_index/genome.txt"
     params:
         extra="",  # optional additional args
     cache: True
@@ -65,8 +63,6 @@ rule download_regulatory_annotation:
         "logs/download_regulatory_annotation.log",
     params:
         release=config["reference"].get("release", "107"),
-    benchmark:
-        "benchmarks/download_regulatory_annotation.txt"
     cache: "omit-software"  # save space and time with between workflow caching (see docs)
     conda:
         "../envs/wget.yaml"
@@ -81,8 +77,6 @@ rule download_repeatmasker_annotation:
         "logs/download_repeatmasker_annotation.log",
     params:
         download_link=config["reference"].get("repeat_masker_download_link", ""),
-    benchmark:
-        "benchmarks/download_repeatmasker_annotation.txt"
     cache: "omit-software"  # save space and time with between workflow caching (see docs)
     conda:
         "../envs/wget.yaml"
@@ -97,8 +91,8 @@ rule download_gene_annotation:
         species=config["reference"]["species"],
         build=config["reference"]["build"],
         release=config["reference"]["release"],
-        flavor="",  # optional, e.g. chr_patch_hapl_scaff, see Ensembl FTP.
-        branch="",  # optional: specify branch
+        flavor=config["reference"].get("flavor", ""),  # optional, e.g. chr_patch_hapl_scaff, see Ensembl FTP.
+        branch=config["reference"].get("branch", ""),  # optional: specify branch
     log:
         "logs/download_gene_annotation.log",
     cache: "omit-software"  # save space and time with between workflow caching (see docs)

workflow/schemas/config.schema.yaml

@@ -34,6 +34,12 @@ properties:
         type: integer
       build:
         type: string
+      flavor:
+        type: string
+      branch:
+        type: string
+      repeat_masker_download_link:
+        type: string
     required:
       - species
       - release
@@ -70,8 +76,11 @@ properties:
           type: number
           minimum: 0.0
           maximum: 1.0
-        local:
-          type: boolean
+        mode:
+          type: array
+          items:
+            type: string
+            enum: ["local-smart", "local-strict", "global-smart", "global-strict"]
         events:
           $ref: "#/definitions/evententry"
           description: "a map of <eventname: event> pairs"