feat: support branches (e.g. plants) in ensembl wrappers for sequence…

…, annotation, and variation download (#546)

bio/reference/ensembl-annotation/test/Snakefile

@@ -6,6 +6,7 @@ rule get_annotation:
         release="87",
         build="GRCh37",
         flavor="",  # optional, e.g. chr_patch_hapl_scaff, see Ensembl FTP.
+        # branch="plants",  # optional: specify branch
     log:
         "logs/get_annotation.log",
     cache: True  # save space and time with between workflow caching (see docs)
@@ -21,6 +22,7 @@ rule get_annotation_gz:
         release="87",
         build="GRCh37",
         flavor="",  # optional, e.g. chr_patch_hapl_scaff, see Ensembl FTP.
+        # branch="plants",  # optional: specify branch
     log:
         "logs/get_annotation.log",
     cache: True  # save space and time with between workflow caching (see docs)

bio/reference/ensembl-annotation/wrapper.py

@@ -24,6 +24,8 @@
 if release >= 81 and build == "GRCh37":
     # use the special grch37 branch for new releases
     branch = "grch37/"
+elif snakemake.params.get("branch"):
+    branch = snakemake.params.branch + "/"
 
 
 flavor = snakemake.params.get("flavor", "")

bio/reference/ensembl-sequence/test/Snakefile

@@ -1,28 +1,30 @@
 rule get_genome:
     output:
-        "refs/genome.fasta"
+        "refs/genome.fasta",
     params:
         species="saccharomyces_cerevisiae",
         datatype="dna",
         build="R64-1-1",
-        release="98"
+        release="98",
     log:
-        "logs/get_genome.log"
+        "logs/get_genome.log",
     cache: True  # save space and time with between workflow caching (see docs)
     wrapper:
         "master/bio/reference/ensembl-sequence"
 
+
 rule get_chromosome:
     output:
-        "refs/chr1.fasta"
+        "refs/chr1.fasta",
     params:
         species="saccharomyces_cerevisiae",
         datatype="dna",
         build="R64-1-1",
         release="101",
-        chromosome="I"
+        chromosome="I",  # optional: restrict to chromosome
+        # branch="plants",  # optional: specify branch
     log:
-        "logs/get_genome.log"
+        "logs/get_genome.log",
     cache: True  # save space and time with between workflow caching (see docs)
     wrapper:
         "master/bio/reference/ensembl-sequence"

bio/reference/ensembl-sequence/wrapper.py

@@ -16,6 +16,8 @@
 if release >= 81 and build == "GRCh37":
     # use the special grch37 branch for new releases
     branch = "grch37/"
+elif snakemake.params.get("branch"):
+    branch = snakemake.params.branch + "/"
 
 log = snakemake.log_fmt_shell(stdout=False, stderr=True)
 

bio/reference/ensembl-variation/test/Snakefile

@@ -11,6 +11,7 @@ rule get_variation:
         build="R64-1-1",
         type="all",  # one of "all", "somatic", "structural_variation"
         # chromosome="21", # optionally constrain to chromosome, only supported for homo_sapiens
+        # branch="plants",  # optional: specify branch
     log:
         "logs/get_variation.log",
     cache: True  # save space and time with between workflow caching (see docs)

bio/reference/ensembl-variation/wrapper.py

@@ -16,14 +16,17 @@
 type = snakemake.params.type
 chromosome = snakemake.params.get("chromosome", "")
 
-if release < 98:
-    print("Ensembl releases <98 are unsupported.", file=open(snakemake.log[0], "w"))
-    exit(1)
 
 branch = ""
 if release >= 81 and build == "GRCh37":
     # use the special grch37 branch for new releases
     branch = "grch37/"
+elif snakemake.params.get("branch"):
+    branch = snakemake.params.branch + "/"
+
+if release < 98 and not branch:
+    print("Ensembl releases <98 are unsupported.", file=open(snakemake.log[0], "w"))
+    exit(1)
 
 log = snakemake.log_fmt_shell(stdout=False, stderr=True)