feat: ensembl reference collections

bio/reference/ensembl-annotation/meta.yaml

@@ -6,3 +6,5 @@ output:
   - Ensemble GTF or GFF3 anotation file
 params:
   - url: URL from where to download cache data (optional; by default is ``ftp://ftp.ensembl.org/pub``)
+  - branch: branch of ftp server to download cache data if required (optional; e.g. "plants")
+  - collection: collection of ftp server to download cache data if required (optional; e.g. "bacteria_0_collection")

bio/reference/ensembl-annotation/test/Snakefile

@@ -6,7 +6,6 @@ rule get_annotation:
         release="105",
         build="GRCh37",
         flavor="",  # optional, e.g. chr_patch_hapl_scaff, see Ensembl FTP.
-        # branch="plants",  # optional: specify branch
     log:
         "logs/get_annotation.log",
     cache: "omit-software"  # save space and time with between workflow caching (see docs)
@@ -22,11 +21,25 @@ rule get_annotation_gz:
         release="105",
         build="GRCh37",
         flavor="",  # optional, e.g. chr_patch_hapl_scaff, see Ensembl FTP.
-        # branch="plants",  # optional: specify branch
     log:
         "logs/get_annotation.log",
+    cache: "omit-software"  # save space and time with between workflow caching (see docs)
+    wrapper:
+        "master/bio/reference/ensembl-annotation"
+
+
+rule get_off_branch_annotation:
+    output:
+        "refs/off_branch_annotation.gtf",
     params:
-        url="http://ftp.ensembl.org/pub",
+        species="bacillus_subtilis_subsp_subtilis_str_168_gca_000009045",
+        release="59", # note latest release varies with url
+        build="ASM904v1",
+        branch="bacteria", # optional for off branch genomes
+        url="ftp://ftp.ensemblgenomes.org/pub/", # optional set ftp server source
+        collection="bacteria_0_collection", # optional set collection source for genome
+    log:
+        "logs/get_annotation.log",
     cache: "omit-software"  # save space and time with between workflow caching (see docs)
     wrapper:
         "master/bio/reference/ensembl-annotation"

bio/reference/ensembl-annotation/wrapper.py

@@ -31,6 +31,9 @@
 elif snakemake.params.get("branch"):
     branch = snakemake.params.branch + "/"
 
+collection = ""
+if snakemake.params.get("collection"):
+    collection = snakemake.params.collection + "/"
 
 flavor = snakemake.params.get("flavor", "")
 if flavor:
@@ -49,7 +52,7 @@
 
 
 url = snakemake.params.get("url", "ftp://ftp.ensembl.org/pub")
-url = f"{url}/{branch}release-{release}/{out_fmt}/{species}/{species.capitalize()}.{build}.{gtf_release}.{flavor}{suffix}"
+url = f"{url}/{branch}release-{release}/{out_fmt}/{collection}{species}/{species.capitalize()}.{build}.{gtf_release}.{flavor}{suffix}"
 
 
 try:

bio/reference/ensembl-sequence/meta.yaml

@@ -6,3 +6,5 @@ output:
   - fasta file
 params:
   - url: URL from where to download cache data (optional; by default is ``ftp://ftp.ensembl.org/pub``)
+  - branch: branch of ftp server to download cache data if required (optional; e.g. "plants")
+  - collection: collection of ftp server to download cache data if required (optional; e.g. "bacteria_0_collection")

bio/reference/ensembl-sequence/test/Snakefile

@@ -22,11 +22,8 @@ rule get_single_chromosome:
         build="R64-1-1",
         release="101",
         chromosome=["II"],  # optional: restrict to one or multiple chromosomes, for multiple see below
-        # branch="plants",  # optional: specify branch
     log:
         "logs/get_genome.log",
-    params:
-        url="http://ftp.ensembl.org/pub",
     cache: "omit-software"  # save space and time with between workflow caching (see docs)
     wrapper:
         "master/bio/reference/ensembl-sequence"
@@ -40,7 +37,24 @@ rule get_multiple_chromosome:
         build="R64-1-1",
         release="101",
         chromosome=["I", "II"],  # optional: restrict to one or multiple chromosomes
-        # branch="plants",  # optional: specify branch
+    log:
+        "logs/get_genome.log",
+    cache: "omit-software"  # save space and time with between workflow caching (see docs)
+    wrapper:
+        "master/bio/reference/ensembl-sequence"
+
+
+rule get_off_branch_genome:
+    output:
+        "refs/off_branch_genome.fasta",
+    params:
+        species="bacillus_subtilis_subsp_subtilis_str_168_gca_000009045",
+        datatype="dna",
+        build="ASM904v1",
+        release="59", # note latest release varies with url
+        branch="bacteria", # optional for off branch genomes
+        url="ftp://ftp.ensemblgenomes.org/pub/", # optional set ftp server source
+        collection="bacteria_0_collection", # optional set collection source for genome
     log:
         "logs/get_genome.log",
     cache: "omit-software"  # save space and time with between workflow caching (see docs)

bio/reference/ensembl-sequence/wrapper.py

@@ -19,11 +19,20 @@
 elif snakemake.params.get("branch"):
     branch = snakemake.params.branch + "/"
 
+collection = ""
+if snakemake.params.get("collection"):
+    collection = snakemake.params.collection + "/"
+
 log = snakemake.log_fmt_shell(stdout=False, stderr=True)
 
-spec = ("{build}" if int(release) > 75 else "{build}.{release}").format(
-    build=build, release=release
-)
+if branch == "" or branch == "grch37/":
+    spec = ("{build}" if int(release) > 75 else "{build}.{release}").format(
+        build=build, release=release
+    )
+else:
+    spec = ("{build}" if int(release) > 30 else "{build}.{release}").format(
+        build=build, release=release
+    )
 
 suffixes = ""
 datatype = snakemake.params.get("datatype", "")
@@ -52,7 +61,7 @@
 
 url = snakemake.params.get("url", "ftp://ftp.ensembl.org/pub")
 spec = spec.format(build=build, release=release)
-url_prefix = f"{url}/{branch}release-{release}/fasta/{species}/{datatype}/{species.capitalize()}.{spec}"
+url_prefix = f"{url}/{branch}release-{release}/fasta/{collection}{species}/{datatype}/{species.capitalize()}.{spec}"
 
 success = False
 for suffix in suffixes:

test.py

@@ -5564,6 +5564,14 @@ def test_ensembl_sequence_chromosomes():
     )
 
 
+@skip_if_not_modified
+def test_ensembl_sequence_off_branch():
+    run(
+        "bio/reference/ensembl-sequence",
+        ["snakemake", "--cores", "1", "refs/off_branch_genome.fasta", "--use-conda", "-F"],
+    )
+
+
 @skip_if_not_modified
 def test_ensembl_sequence_chromosome_old_release():
     run(
@@ -5597,6 +5605,14 @@ def test_ensembl_annotation_gtf_gz():
     )
 
 
+@skip_if_not_modified
+def test_ensembl_off_branch_annotation_gtf():
+    run(
+        "bio/reference/ensembl-annotation",
+        ["snakemake", "--cores", "1", "refs/off_branch_annotation.gtf", "--use-conda", "-F"],
+    )
+
+
 @skip_if_not_modified
 def test_ensembl_variation():
     run(