Merge pull request #31 from multimeric/ci-fixes

CI fixes
multimeric · May 18, 2024 · 7a97b8a · 7a97b8a
2 parents fed6ef3 + ad330f4
commit 7a97b8a
Show file tree

Hide file tree

Showing 8 changed files with 2,036 additions and 1,456 deletions.
diff --git a/poetry.lock b/poetry.lock
diff --git a/test/test_dataset/test_request.py b/test/test_dataset/test_request.py
@@ -87,7 +87,12 @@ def test_main_example():
         fields=["length", "gene_names"],
     ).each_record():
         assert isinstance(record, dict)
-        assert set(record.keys()) == {"primaryAccession", "genes", "sequence"}
+        assert set(record.keys()) == {
+            "primaryAccession",
+            "genes",
+            "sequence",
+            "entryType",
+        }
         assert record["sequence"]["length"] > 5000
 
 
@@ -104,7 +109,7 @@ def test_date_field():
             fields=["date_created", "protein_name"],
         ).each_record()
     )
-    assert len(records) == 544
+    assert len(records) == 542
 
 
 def test_uniref():

diff --git a/test/test_dataset/test_validate_fields.py b/test/test_dataset/test_validate_fields.py
@@ -48,4 +48,4 @@ def test_valid_return_fields(client: Type[DatasetClient]):
             ).each_response()
         )
     )
-    response.raise_for_status()
+    assert response.status_code == 200, response.json()
diff --git a/unipressed/dataset/generated_types/database.py b/unipressed/dataset/generated_types/database.py
@@ -27,7 +27,7 @@ class DatabaseQueryDict(TypedDict):
     "pubmed_id",
     "doi_id",
     "link_type",
-    "server",
+    "servers",
     "dbUrl",
     "category",
     "statistics",

diff --git a/unipressed/dataset/generated_types/keywords.py b/unipressed/dataset/generated_types/keywords.py
@@ -5,6 +5,19 @@
 
 from typing_extensions import Literal, NotRequired, TypeAlias, TypedDict
 
+Category: TypeAlias = Literal[
+    "technical_term",
+    "ptm",
+    "molecular_function",
+    "ligand",
+    "domain",
+    "disease",
+    "developmental_stage",
+    "coding_sequence_diversity",
+    "cellular_component",
+    "biological_process",
+]
+
 
 class KeywordsQueryDict(TypedDict):
     and_: NotRequired[Iterable["KeywordsQuery"]]
@@ -17,7 +30,7 @@ class KeywordsQueryDict(TypedDict):
     "Name\ne.g. 2Fe-2S"
     keyword_id: NotRequired[str]
     "Keyword [AC]\ne.g. KW-0001"
-    category: NotRequired[str]
+    category: NotRequired[Category]
     "Category\ne.g. Domain\n* technical_term: Technical term [KW-9990]\n* ptm: PTM [KW-9991]\n* molecular_function: Molecular function [KW-9992]\n* ligand: Ligand [KW-9993]\n* domain: Domain [KW-9994]\n* disease: Disease [KW-9995]\n* developmental_stage: Developmental stage [KW-9996]\n* coding_sequence_diversity: Coding sequence diversity [KW-9997]\n* cellular_component: Cellular component [KW-9998]\n* biological_process: Biological process [KW-9999]"
 
 

diff --git a/unipressed/dataset/generated_types/taxonomy.py b/unipressed/dataset/generated_types/taxonomy.py
@@ -62,7 +62,7 @@ class TaxonomyQueryDict(TypedDict):
     host: NotRequired[int]
     "Virus host\ne.g. 85621"
     linked: NotRequired[bool]
-    "With external info\ne.g. true\n* true: Yes\n* false: No"
+    "With external info\ne.g. true"
     parent: NotRequired[str]
     "Parent\ne.g. 9606"
     ancestor: NotRequired[str]

diff --git a/unipressed/dataset/generated_types/uniparc.py b/unipressed/dataset/generated_types/uniparc.py
@@ -5,6 +5,80 @@
 
 from typing_extensions import Literal, NotRequired, TypeAlias, TypedDict
 
+Database: TypeAlias = Literal[
+    "EnsemblBacteria",
+    "EnsemblFungi",
+    "EnsemblMetazoa",
+    "EnsemblPlants",
+    "EnsemblProtists",
+    "embl-cds",
+    "EMBL_CON",
+    "EMBL_TPA",
+    "EMBL_TSA",
+    "EMBLWGS",
+    "Ensembl",
+    "EnsemblRapid",
+    "EPO",
+    "FlyBase",
+    "FusionGDB",
+    "H-InvDB",
+    "IPI",
+    "JPO",
+    "KIPO",
+    "PATRIC",
+    "PDB",
+    "PIR",
+    "PIRARC",
+    "PRF",
+    "RefSeq",
+    "REMTREMBL",
+    "SEED",
+    "SGD",
+    "UniProt",
+    "isoforms",
+    "TAIR",
+    "TREMBLNEW",
+    "TREMBL_VARSPLIC",
+    "TROME",
+    "UNIMES",
+    "USPTO",
+    "VectorBase",
+    "VEGA",
+    "WBParaSite",
+    "WormBase",
+]
+Active: TypeAlias = Literal[
+    "EnsemblBacteria",
+    "EnsemblFungi",
+    "EnsemblMetazoa",
+    "EnsemblPlants",
+    "EnsemblProtists",
+    "embl-cds",
+    "EMBL_CON",
+    "EMBL_TSA",
+    "EMBLWGS",
+    "Ensembl",
+    "EnsemblRapid",
+    "EPO",
+    "FlyBase",
+    "FusionGDB",
+    "JPO",
+    "KIPO",
+    "PATRIC",
+    "PDB",
+    "RefSeq",
+    "SEED",
+    "SGD",
+    "UniProt",
+    "isoforms",
+    "TAIR",
+    "TROME",
+    "USPTO",
+    "VEGA",
+    "WBParaSite",
+    "WormBase",
+]
+
 
 class UniparcQueryDict(TypedDict):
     and_: NotRequired[Iterable["UniparcQuery"]]
@@ -22,19 +96,19 @@ class UniparcQueryDict(TypedDict):
     upid: NotRequired[str]
     "Proteome ID\ne.g. UP123456789"
     taxonomy_name: NotRequired[str]
-    "Taxonomy [OC]\ne.g. sample name"
+    "Taxonomy [OC]\ne.g. Human"
     taxonomy_id: NotRequired[str]
     "Taxonomy id"
     gene: NotRequired[str]
-    "Gene name [GN]\ne.g. sample gene"
+    "Gene name [GN]\ne.g. PROZ"
     protein: NotRequired[str]
-    "Protein name\ne.g. sample protein"
-    database: NotRequired[str]
-    "Database\ne.g. sample database\n* EnsemblBacteria: EnsemblBacteria\n* EnsemblFungi: EnsemblFungi\n* EnsemblMetazoa: EnsemblMetazoa\n* EnsemblPlants: EnsemblPlants\n* EnsemblProtists: EnsemblProtists\n* embl-cds: EMBL CDS\n* EMBL_CON: EMBL_CON\n* EMBL_TPA: EMBL_TPA\n* EMBL_TSA: EMBL_TSA\n* EMBLWGS: EMBLWGS\n* Ensembl: Ensembl\n* EnsemblRapid: EnsemblRapid\n* EPO: EPO\n* FlyBase: FlyBase\n* FusionGDB: FusionGDB\n* H-InvDB: H-InvDB\n* IPI: IPI\n* JPO: JPO\n* KIPO: KIPO\n* PATRIC: PATRIC\n* PDB: PDB\n* PIR: PIR\n* PIRARC: PIRARC\n* PRF: PRF\n* RefSeq: RefSeq\n* REMTREMBL: REMTREMBL\n* SEED: SEED\n* SGD: SGD\n* UniProt: UniProtKB\n* isoforms: UniProtKB/Swiss-Prot isoforms\n* TAIR: TAIR\n* TREMBLNEW: TREMBLNEW\n* TREMBL_VARSPLIC: TREMBL_VARSPLIC\n* TROME: TROME\n* UNIMES: UNIMES\n* USPTO: USPTO\n* VectorBase: VectorBase\n* VEGA: VEGA\n* WBParaSite: WBParaSite\n* WormBase: WormBase"
-    active: NotRequired[str]
-    "Active\ne.g. sample active\n* EnsemblBacteria: EnsemblBacteria\n* EnsemblFungi: EnsemblFungi\n* EnsemblMetazoa: EnsemblMetazoa\n* EnsemblPlants: EnsemblPlants\n* EnsemblProtists: EnsemblProtists\n* embl-cds: EMBL CDS\n* EMBL_CON: EMBL_CON\n* EMBL_TSA: EMBL_TSA\n* EMBLWGS: EMBLWGS\n* Ensembl: Ensembl\n* EnsemblRapid: EnsemblRapid\n* EPO: EPO\n* FlyBase: FlyBase\n* FusionGDB: FusionGDB\n* JPO: JPO\n* KIPO: KIPO\n* PATRIC: PATRIC\n* PDB: PDB\n* RefSeq: RefSeq\n* SEED: SEED\n* SGD: SGD\n* UniProt: UniProtKB\n* isoforms: UniProtKB/Swiss-Prot isoforms\n* TAIR: TAIR\n* TROME: TROME\n* USPTO: USPTO\n* VEGA: VEGA\n* WBParaSite: WBParaSite\n* WormBase: WormBase"
+    "Protein name\ne.g. Protein Z"
+    database: NotRequired[Database]
+    "Database\ne.g. Gene3D\n* EnsemblBacteria: EnsemblBacteria\n* EnsemblFungi: EnsemblFungi\n* EnsemblMetazoa: EnsemblMetazoa\n* EnsemblPlants: EnsemblPlants\n* EnsemblProtists: EnsemblProtists\n* embl-cds: EMBL CDS\n* EMBL_CON: EMBL_CON\n* EMBL_TPA: EMBL_TPA\n* EMBL_TSA: EMBL_TSA\n* EMBLWGS: EMBLWGS\n* Ensembl: Ensembl\n* EnsemblRapid: EnsemblRapid\n* EPO: EPO\n* FlyBase: FlyBase\n* FusionGDB: FusionGDB\n* H-InvDB: H-InvDB\n* IPI: IPI\n* JPO: JPO\n* KIPO: KIPO\n* PATRIC: PATRIC\n* PDB: PDB\n* PIR: PIR\n* PIRARC: PIRARC\n* PRF: PRF\n* RefSeq: RefSeq\n* REMTREMBL: REMTREMBL\n* SEED: SEED\n* SGD: SGD\n* UniProt: UniProtKB\n* isoforms: UniProtKB/Swiss-Prot isoforms\n* TAIR: TAIR\n* TREMBLNEW: TREMBLNEW\n* TREMBL_VARSPLIC: TREMBL_VARSPLIC\n* TROME: TROME\n* UNIMES: UNIMES\n* USPTO: USPTO\n* VectorBase: VectorBase\n* VEGA: VEGA\n* WBParaSite: WBParaSite\n* WormBase: WormBase"
+    active: NotRequired[Active]
+    "Active\ne.g. Gene3D\n* EnsemblBacteria: EnsemblBacteria\n* EnsemblFungi: EnsemblFungi\n* EnsemblMetazoa: EnsemblMetazoa\n* EnsemblPlants: EnsemblPlants\n* EnsemblProtists: EnsemblProtists\n* embl-cds: EMBL CDS\n* EMBL_CON: EMBL_CON\n* EMBL_TSA: EMBL_TSA\n* EMBLWGS: EMBLWGS\n* Ensembl: Ensembl\n* EnsemblRapid: EnsemblRapid\n* EPO: EPO\n* FlyBase: FlyBase\n* FusionGDB: FusionGDB\n* JPO: JPO\n* KIPO: KIPO\n* PATRIC: PATRIC\n* PDB: PDB\n* RefSeq: RefSeq\n* SEED: SEED\n* SGD: SGD\n* UniProt: UniProtKB\n* isoforms: UniProtKB/Swiss-Prot isoforms\n* TAIR: TAIR\n* TROME: TROME\n* USPTO: USPTO\n* VEGA: VEGA\n* WBParaSite: WBParaSite\n* WormBase: WormBase"
     checksum: NotRequired[str]
-    "Checksum (CRC64/MD5)\ne.g. sample checksum"
+    "Checksum (CRC64/MD5)\ne.g. B8824CE1ECAEEEAE"
     length: NotRequired[
         tuple[
             Union[
@@ -75,7 +149,7 @@ class UniparcQueryDict(TypedDict):
     "SFLD",
     "SMART",
     "SUPFAM",
-    "TIGRFAMs",
+    "NCBIfam",
 ]
 UniparcFields: TypeAlias = Literal[
     UniparcNamesTaxonomy,

diff --git a/unipressed/dataset/generated_types/uniprotkb.py b/unipressed/dataset/generated_types/uniprotkb.py
@@ -8,6 +8,7 @@
 Existence: TypeAlias = Literal["1", "2", "3", "4", "5"]
 Organelle: TypeAlias = Literal[
     "mitochondrion",
+    "plasmid",
     "plastid",
     "chloroplast",
     "cyanelle",
@@ -245,7 +246,7 @@ class UniprotkbQueryDict(TypedDict):
     cc_subunit_exp: NotRequired[str]
     "Cc subunit exp\ne.g. homodimer"
     structure_3d: NotRequired[bool]
-    "3D Structure\ne.g. true\n* true: Yes\n* false: No"
+    "3D Structure\ne.g. true"
     ft_secstruct: NotRequired[str]
     "Ft secstruct\ne.g. *"
     ft_secstruct_exp: NotRequired[str]
@@ -375,11 +376,11 @@ class UniprotkbQueryDict(TypedDict):
     ft_positional_exp: NotRequired[str]
     "Ft positional exp\ne.g. colorectal"
     fragment: NotRequired[bool]
-    "Fragment\ne.g. true\n* true: Yes\n* false: No"
+    "Fragment\ne.g. true"
     organelle: NotRequired[Organelle]
-    "Encoded in\ne.g. mitochondrion\n* mitochondrion: Mitochondrion\n* plastid: Plastid\n* chloroplast: Chloroplast\n* cyanelle: Cyanelle\n* apicoplast: Apicoplast\n* organellar chromatophore: Organellar chromatophore\n* non-photosynthetic plastid: Non-photosynthetic plastid\n* nucleomorph: Nucleomorph\n* hydrogenosome: Hydrogenosome"
+    "Encoded in\ne.g. mitochondrion\n* mitochondrion: Mitochondrion\n* plasmid: Plasmid\n* plastid: Plastid\n* chloroplast: Chloroplast\n* cyanelle: Cyanelle\n* apicoplast: Apicoplast\n* organellar chromatophore: Organellar chromatophore\n* non-photosynthetic plastid: Non-photosynthetic plastid\n* nucleomorph: Nucleomorph\n* hydrogenosome: Hydrogenosome"
     precursor: NotRequired[bool]
-    "Precursor\ne.g. true\n* true: Yes\n* false: No"
+    "Precursor\ne.g. true"
     tissue: NotRequired[str]
     "Tissue\ne.g. head"
     strain: NotRequired[str]
@@ -403,9 +404,9 @@ class UniprotkbQueryDict(TypedDict):
     ft_coiled_exp: NotRequired[str]
     "Ft coiled exp\ne.g. *"
     ft_compbias: NotRequired[str]
-    "Ft compbias\ne.g. glu-rich"
+    "Ft compbias\ne.g. basic residues"
     ft_compbias_exp: NotRequired[str]
-    "Ft compbias exp\ne.g. glu-rich"
+    "Ft compbias exp\ne.g. basic residues"
     ft_motif: NotRequired[str]
     "Ft motif\ne.g. motif"
     ft_motif_exp: NotRequired[str]
@@ -561,9 +562,9 @@ class UniprotkbQueryDict(TypedDict):
     scope: NotRequired[str]
     "Cited for\ne.g. microtubule"
     reviewed: NotRequired[bool]
-    "Reviewed\ne.g. true\n* true: Yes\n* false: No"
+    "Reviewed\ne.g. true"
     active: NotRequired[bool]
-    "Active\ne.g. true\n* true: Yes\n* false: No"
+    "Active\ne.g. true"
     uniref_cluster_50: NotRequired[str]
     "UniRef50\ne.g. UniRef50_P05067"
     uniref_cluster_90: NotRequired[str]
@@ -677,7 +678,7 @@ class UniprotkbQueryDict(TypedDict):
 UniprotkbStructure: TypeAlias = Literal[
     "structure_3d", "ft_strand", "ft_helix", "ft_turn"
 ]
-UniprotkbPublications: TypeAlias = Literal["lit_pubmed_id",]
+UniprotkbPublications: TypeAlias = Literal["lit_pubmed_id", "lit_doi_id"]
 UniprotkbDateOf: TypeAlias = Literal[
     "date_created", "date_modified", "date_sequence_modified", "version"
 ]
@@ -690,14 +691,16 @@ class UniprotkbQueryDict(TypedDict):
     "protein_families",
     "ft_region",
     "ft_repeat",
+    "cc_similarity",
     "ft_zn_fing",
 ]
 UniprotkbSequence: TypeAlias = Literal[
-    "xref_ccds", "xref_embl", "xref_pir", "xref_refseq"
+    "xref_ccds", "xref_embl", "xref_generif", "xref_pir", "xref_refseq"
 ]
 UniprotkbThreedStructure: TypeAlias = Literal[
     "xref_alphafolddb",
     "xref_bmrb",
+    "xref_emdb",
     "xref_pcddb",
     "xref_pdb",
     "xref_pdbsum",
@@ -749,7 +752,7 @@ class UniprotkbQueryDict(TypedDict):
     "xref_iptmnet",
 ]
 UniprotkbPolymorphismAndMutation: TypeAlias = Literal[
-    "xref_biomuta", "xref_dmdm", "xref_dbsnp"
+    "xref_alzforum", "xref_biomuta", "xref_dmdm", "xref_dbsnp"
 ]
 UniprotkbTwodGel: TypeAlias = Literal[
     "xref_compluyeast-2dpage",
@@ -770,6 +773,7 @@ class UniprotkbQueryDict(TypedDict):
     "xref_peptideatlas",
     "xref_promex",
     "xref_proteomicsdb",
+    "xref_pumba",
     "xref_topdownproteomics",
     "xref_jpost",
 ]
@@ -791,7 +795,6 @@ class UniprotkbQueryDict(TypedDict):
     "xref_ucsc",
     "xref_vectorbase",
     "xref_wbparasite",
-    "xref_wbparasitetranscriptprotein",
 ]
 UniprotkbOrganismSpecific: TypeAlias = Literal[
     "xref_agr",
@@ -807,6 +810,8 @@ class UniprotkbQueryDict(TypedDict):
     "xref_genereviews",
     "xref_hgnc",
     "xref_hpa",
+    "xref_ic4r",
+    "xref_japonicusdb",
     "xref_legiolist",
     "xref_leproma",
     "xref_mgi",
@@ -860,10 +865,14 @@ class UniprotkbQueryDict(TypedDict):
     "xref_evolutionarytrace",
     "xref_genewiki",
     "xref_genomernai",
+    "xref_orcid",
+    "xref_pgenn",
     "xref_phi-base",
     "xref_pro",
     "xref_pharos",
+    "xref_pubtator",
     "xref_rnact",
+    "xref_emind",
 ]
 UniprotkbGeneExpression: TypeAlias = Literal[
     "xref_bgee",
@@ -879,16 +888,15 @@ class UniprotkbQueryDict(TypedDict):
     "xref_hamap",
     "xref_ideal",
     "xref_interpro",
+    "xref_ncbifam",
     "xref_panther",
     "xref_pirsf",
     "xref_prints",
     "xref_prosite",
     "xref_pfam",
-    "xref_prodom",
     "xref_sfld",
     "xref_smart",
     "xref_supfam",
-    "xref_tigrfams",
 ]
 UniprotkbFields: TypeAlias = Literal[
     UniprotkbNamesTaxonomy,