Merge pull request #1042 from andrewkern/dros_annot

merging the annotation side generalization. 

closes #618, resolves #562, close #659

maintenance/annotation_maint.py

@@ -96,18 +96,18 @@ def download_process_annotations():
         if spc.annotations:
             for an in spc.annotations:
                 CHROM_IDS = [chrom.id for chrom in spc.genome.chromosomes]
-                logger.info(f"Downloading GFF file {spc.id}")
+                logger.info(f"Downloading GFF file {an.id}")
                 gff = get_gff_recarray(an.url, an.gff_sha256)
-                logger.info("extracting exons {spc.id}")
-                # this is fragile-- is ensembl_havana always a feature?
+                logger.info(f"extracting annotations {an.id}")
                 exons = gff[
                     np.where(
                         np.logical_and(
-                            gff.source == "ensembl_havana", gff.type == "exon"
+                            gff.source == an.annotation_source,
+                            gff.type == an.annotation_type,
                         )
                     )
                 ]
-                logger.info(f"merging overlapping regions {spc.id}")
+                logger.info(f"merging overlapping regions {an.id}")
                 # create zarr store and zarr root
                 spc_name_path = os.path.join(annot_path, spc.id)
                 os.makedirs(spc_name_path, exist_ok=True)

stdpopsim/annotations.py

@@ -13,7 +13,7 @@
 @attr.s(kw_only=True)
 class Annotation:
     """
-    Class representing a GFF3 annotation file.
+    Class representing an annotation track.
 
     :ivar str ~.id: String that uniquely identifies the annotation.
     :ivar species: The species to which this annotation applies.
@@ -37,6 +37,8 @@ class Annotation:
     description = attr.ib()
     citations = attr.ib(factory=list)
     file_pattern = attr.ib()
+    annotation_source = attr.ib()
+    annotation_type = attr.ib()
 
     def __attrs_post_init__(self):
         self._cache = stdpopsim.CachedData(

stdpopsim/catalog/DroMel/__init__.py

@@ -5,3 +5,4 @@
 from . import genetic_maps  # noqa: F401
 from . import demographic_models  # noqa: F401
 from . import dfes  # noqa: F401
+from . import annotations  # noqa: F401

stdpopsim/catalog/DroMel/annotations.py

@@ -0,0 +1,61 @@
+import stdpopsim
+
+_species = stdpopsim.get_species("DroMel")
+# exons
+_an = stdpopsim.Annotation(
+    species=_species,
+    id="FlyBase_BDGP6.32.51_exons",
+    description="FlyBase exon annotations on BDGP6",
+    url=(
+        "http://ftp.ebi.ac.uk/ensemblgenomes/pub/current/metazoa/"
+        "gff3/drosophila_melanogaster/"
+        "Drosophila_melanogaster.BDGP6.32.51.gff3.gz"
+    ),
+    gff_sha256="d882d9a2af1c090ad69b4c81e54b809506f7a8d5fdd90597c6ed05c79ad502bc",
+    intervals_url=(
+        "https://stdpopsim.s3-us-west-2.amazonaws.com/"
+        "annotations/DroMel/FlyBase_BDGP6.32.51_exons.tar.gz"
+    ),
+    intervals_sha256="680ee9b0a565f85c561cd3672927cb2bc8649405a750532f91e1717b3ed8b993",
+    citations=[
+        stdpopsim.Citation(
+            year=2014,
+            author="Hoskins et al",
+            doi="https://doi.org/10.1101/gr.185579.114",
+            reasons={stdpopsim.CiteReason.ANNOTATION},
+        )
+    ],
+    file_pattern="flybase_exons_{id}.txt",
+    annotation_source="FlyBase",
+    annotation_type="exon",
+)
+_species.add_annotations(_an)
+
+_an2 = stdpopsim.Annotation(
+    species=_species,
+    id="FlyBase_BDGP6.32.51_CDS",
+    description="FlyBase CDS annotations on BDGP6",
+    url=(
+        "http://ftp.ebi.ac.uk/ensemblgenomes/pub/current/metazoa/"
+        "gff3/drosophila_melanogaster/"
+        "Drosophila_melanogaster.BDGP6.32.51.gff3.gz"
+    ),
+    gff_sha256="d882d9a2af1c090ad69b4c81e54b809506f7a8d5fdd90597c6ed05c79ad502bc",
+    intervals_url=(
+        "https://stdpopsim.s3-us-west-2.amazonaws.com/"
+        "annotations/DroMel/FlyBase_BDGP6.32.51_CDS.tar.gz"
+    ),
+    intervals_sha256="5f202f454e7d0051f863a5146e775c85e4f3e39bab434cc96701d46026eb7364",
+    citations=[
+        stdpopsim.Citation(
+            year=2014,
+            author="Hoskins et al",
+            doi="https://doi.org/10.1101/gr.185579.114",
+            reasons={stdpopsim.CiteReason.ANNOTATION},
+        )
+    ],
+    file_pattern="flybase_CDS_{id}.txt",
+    annotation_source="FlyBase",
+    annotation_type="CDS",
+)
+_species.add_annotations(_an2)

stdpopsim/catalog/HomSap/annotations.py

@@ -4,18 +4,18 @@
 
 _an = stdpopsim.Annotation(
     species=_species,
-    id="Ensembl_GRCh38_104_gff3",
-    description="Ensembl GFF3 annotations on GRCh38",
+    id="ensembl_havana_104_exons",
+    description="Ensembl Havana exon annotations on GRCh38",
     url=(
         "ftp://ftp.ensembl.org/pub/release-104/"
         "gff3/homo_sapiens/Homo_sapiens.GRCh38.104.gff3.gz"
     ),
     gff_sha256="313ad46bd4af78b45b9f5d8407bbcbd3f87f4be0747060e84b3b5eb931530ec1",
     intervals_url=(
         "https://stdpopsim.s3-us-west-2.amazonaws.com/"
-        "annotations/HomSap/Ensembl_GRCh38_104_gff3.tar.gz"
+        "annotations/HomSap/ensembl_havana_104_exons.tar.gz"
     ),
-    intervals_sha256="b0e864ec87274f3084e0d93161c8ed959b845c97a42cbad5bfe33f54c862716d",
+    intervals_sha256="5c356d092b31fa40bfce434994de276e9040ed9a80fc047a5e3b94410157f1cf",
     citations=[
         stdpopsim.Citation(
             year=2018,
@@ -25,5 +25,36 @@
         )
     ],
     file_pattern="ensembl_havana_exons_{id}.txt",
+    annotation_source="ensembl_havana",
+    annotation_type="exon",
 )
 _species.add_annotations(_an)
+
+# add CDS
+_an2 = stdpopsim.Annotation(
+    species=_species,
+    id="ensembl_havana_104_CDS",
+    description="Ensembl Havana CDS annotations on GRCh38",
+    url=(
+        "ftp://ftp.ensembl.org/pub/release-104/"
+        "gff3/homo_sapiens/Homo_sapiens.GRCh38.104.gff3.gz"
+    ),
+    gff_sha256="313ad46bd4af78b45b9f5d8407bbcbd3f87f4be0747060e84b3b5eb931530ec1",
+    intervals_url=(
+        "https://stdpopsim.s3-us-west-2.amazonaws.com/"
+        "annotations/HomSap/ensembl_havana_104_CDS.tar.gz"
+    ),
+    intervals_sha256="24b36e6f88a6d995ecaee7ef965baa4d2ea850058ee2f5084efdbb0ea47f1c8e",
+    citations=[
+        stdpopsim.Citation(
+            year=2018,
+            author="Hunt et al",
+            doi="https://doi.org/10.1093/database/bay119",
+            reasons={stdpopsim.CiteReason.ANNOTATION},
+        )
+    ],
+    file_pattern="ensembl_havana_CDS_{id}.txt",
+    annotation_source="ensembl_havana",
+    annotation_type="CDS",
+)
+_species.add_annotations(_an2)