Merge pull request #111 from Ensembl/chore/add_stats_prefix

Add `.stats` Prefix

protos/ensembl/production/metadata/grpc/ensembl_metadata.proto

@@ -248,7 +248,7 @@ message AttributesInfo {
   string genebuild_method = 1;
   string genebuild_method_display = 2;
   string genebuild_last_geneset_update = 3;
-  string genebuild_version = 4;
+  string genebuild_provider_version = 4;
   string genebuild_provider_name = 5;
   string genebuild_provider_url = 6;
   string genebuild_sample_gene = 7;
@@ -383,15 +383,6 @@ message GenomeUUIDRequest {
   double release_version = 3; // Optional
 }
 
-/*
-Genome keyword filter.
-If release_version is not given, the current version is used.
- */
-message GenomeByKeywordRequest {
-  string keyword = 1; // Mandatory
-  double release_version = 2; // Optional
-}
-
 /*
 Genome specific keyword filter.
 If release_version is not given, the current version is used.

src/ensembl/production/metadata/grpc/client_examples.py

@@ -426,10 +426,10 @@ def get_datasets_attributes_values_by_genome_uuid(stub):
     request3 = DatasetAttributesValuesRequest(
         genome_uuid="a73351f7-93e7-11ec-a39d-005056b38ce3",
         dataset_type="homologies",
-        attribute_name=["compara.homology_coverage"]
+        attribute_name=["compara.stats.homology_coverage"]
     )
     attributes3 = stub.GetAttributesValuesByUUID(request3)
-    print("**** Dataset Attributes Values: By genome_uuid, dataset_type='homologies' and attribute_name=['compara.homology_coverage'] ****")
+    print("**** Dataset Attributes Values: By genome_uuid, dataset_type='homologies' and attribute_name=['compara.stats.homology_coverage'] ****")
     print(attributes3)
 
     request4 = DatasetAttributesValuesRequest(

src/ensembl/production/metadata/grpc/protobuf_msg_factory.py

@@ -201,7 +201,7 @@ def create_attributes_info(data=None):
         genebuild_method=required_attributes["genebuild.method"],
         genebuild_method_display=required_attributes["genebuild.method_display"],
         genebuild_last_geneset_update=required_attributes["genebuild.last_geneset_update"],
-        genebuild_version=required_attributes["genebuild.provider_version"],
+        genebuild_provider_version=required_attributes["genebuild.provider_version"],
         genebuild_provider_name=required_attributes["genebuild.provider_name"],
         genebuild_provider_url=required_attributes["genebuild.provider_url"],
         genebuild_sample_gene=required_attributes["genebuild.sample_gene"],
@@ -399,7 +399,7 @@ def populate_dataset_info(data=None):
             dataset_source_type=ds_item.dataset.dataset_source.type if ds_item.dataset.dataset_source else "",
             dataset_type_name=ds_item.dataset.dataset_type.name,
             release_version=float(data.release.version) if data.release.version else None,
-            release_date=datetime.strftime(data.release.release_date, "%m/%d/%Y"),
+            release_date=datetime.strftime(data.release.release_date, "%m/%d/%Y") if data.release.release_date else None,
             release_type=data.release.release_type,
         )
         ds_obj_list.append(ds_info)

src/tests/databases/core_7/meta.txt

@@ -22,4 +22,4 @@
 24	1	genebuild.start_date	2023-08-Ensembl
 25	1	genebuild.havana_datafreeze_date	test2
 26	\N	schema_version	110
-27	1	assembly.total_coding_sequence_length	8989
+27	1	assembly.stats.total_coding_sequence_length	8989

src/tests/databases/core_9/meta.txt

@@ -20,6 +20,6 @@
 20	1	strain.type	test
 23	1	genome.genome_uuid	a733550b-93e7-11ec-a39d-005056b38ce3
 24	1	genebuild.havana_datafreeze_date	test2
-25	1	assembly.total_genome_length	546
+25	1	assembly.stats.total_genome_length	546
 26	1	genebuild.start_date	2023-07-Ensembl
 27	\N	schema_version	110

src/tests/databases/ensembl_genome_metadata/attribute.txt

@@ -1,101 +1,101 @@
 1	assembly.accession	assembly.accession	assembly.accession	string
-2	assembly.chromosomes	Chromosomes or plasmids	Number of structures in cells containing DNA	integer
-3	assembly.component_sequences	Component sequences	Part of the primary sequences in assembly	integer
-4	assembly.contig_n50	Contig N50	Median size of contigs in a genome assembly	bp
+2	assembly.stats.chromosomes	Chromosomes or plasmids	Number of structures in cells containing DNA	integer
+3	assembly.stats.component_sequences	Component sequences	Part of the primary sequences in assembly	integer
+4	assembly.stats.contig_n50	Contig N50	Median size of contigs in a genome assembly	bp
 5	assembly.date	assembly.date	assembly.date	string
 6	assembly.default	assembly.default	assembly.default	string
-7	assembly.gc_percentage	Average GC content	Percentage of nucleotides in DNA that are G or C	percent
+7	assembly.stats.gc_percentage	Average GC content	Percentage of nucleotides in DNA that are G or C	percent
 8	assembly.is_reference	assembly.is_reference	assembly.is_reference	string
 9	assembly.level	assembly.level	assembly.level	string
 10	assembly.mapping	assembly.mapping	assembly.mapping	string
 11	assembly.name	assembly.name	assembly.name	string
 12	assembly.provider_name	assembly.provider_name	assembly.provider_name	string
 13	assembly.provider_url	assembly.provider_url	assembly.provider_url	string
-14	assembly.spanned_gaps	Spanned gaps	Number of gaps covered by sequencing reads	integer
+14	assembly.stats.spanned_gaps	Spanned gaps	Number of gaps covered by sequencing reads	integer
 15	assembly.tolid	assembly.tolid	assembly.tolid	string
-16	assembly.toplevel_sequences	Top level sequences	Primary sequences in a genome assembly	integer
-17	assembly.total_coding_sequence_length	Total coding sequence length	Total length of all coding sequences	bp
-18	assembly.total_gap_length	Total gap length	Total length of all gaps in a genome assembly	bp
-19	assembly.total_genome_length	Total genome length	Total length of all genomic sequences	bp
+16	assembly.stats.toplevel_sequences	Top level sequences	Primary sequences in a genome assembly	integer
+17	assembly.stats.total_coding_sequence_length	Total coding sequence length	Total length of all coding sequences	bp
+18	assembly.stats.total_gap_length	Total gap length	Total length of all gaps in a genome assembly	bp
+19	assembly.stats.total_genome_length	Total genome length	Total length of all genomic sequences	bp
 20	assembly.ucsc_alias	assembly.ucsc_alias	assembly.ucsc_alias	string
-21	genebuild.average_cds_length	Average CDS length	Average length of coding sequences	float
+21	genebuild.stats.average_cds_length	Average CDS length	Average length of coding sequences	float
 22	genebuild.average_coding_exons_per_coding_gene	Average coding exons per coding gene	Average coding exons per coding gene	string
-23	genebuild.average_coding_exons_per_transcript	Average coding exons per transcript	Average coding exons per coding transcript	float
-24	genebuild.average_coding_exon_length	Average exon length per coding gene	Average length of coding exons	bp
-25	genebuild.average_exon_length	Average exon length	Average length of exons	bp
-26	genebuild.average_genomic_span	Average coding genomic span	Average length of all genomic regions	bp
-27	genebuild.average_intron_length	Average intron length	Average intron length per coding gene	bp
+23	genebuild.stats.average_coding_exons_per_transcript	Average coding exons per transcript	Average coding exons per coding transcript	float
+24	genebuild.stats.average_coding_exon_length	Average exon length per coding gene	Average length of coding exons	bp
+25	genebuild.stats.average_exon_length	Average exon length	Average length of exons	bp
+26	genebuild.stats.average_genomic_span	Average coding genomic span	Average length of all genomic regions	bp
+27	genebuild.stats.average_intron_length	Average intron length	Average intron length per coding gene	bp
 28	genebuild.average_sequence_legth	Average coding sequence length	Average length of sequences in genome	bp
-29	genebuild.coding_genes	Coding genes	Genes that code for proteins	integer
-30	genebuild.coding_transcripts	Coding transcripts	Transcripts that code for proteins	integer
-31	genebuild.coding_transcripts_per_gene	Average coding transcripts per gene	Average coding transcripts per gene	float
+29	genebuild.stats.coding_genes	Coding genes	Genes that code for proteins	integer
+30	genebuild.stats.coding_transcripts	Coding transcripts	Transcripts that code for proteins	integer
+31	genebuild.stats.coding_transcripts_per_gene	Average coding transcripts per gene	Average coding transcripts per gene	float
 32	genebuild.hash	genebuild.hash	genebuild.hash	string
 33	genebuild.initial_release_date	genebuild.initial_release_date	genebuild.initial_release_date	string
 34	genebuild.last_geneset_update	genebuild.last_geneset_update	genebuild.last_geneset_update	string
 35	genebuild.level	genebuild.level	genebuild.level	string
-36	genebuild.longest_gene_length	Longest coding gene	Length of longest gene	bp
+36	genebuild.stats.longest_gene_length	Longest coding gene	Length of longest gene	bp
 37	genebuild.method	genebuild.method	genebuild.method	string
 38	genebuild.method_display	genebuild.method_display	genebuild.method_display	string
-39	genebuild.nc_average_exons_per_transcript	Average exons per non-coding transcript	Mean exon count per transcript	float
-40	genebuild.nc_average_exon_length	Average exon length per non-coding transcript	Mean exon length	bp
-41	genebuild.nc_average_genomic_span	Average non-coding genomic span	Mean length of all genomic regions	bp
-42	genebuild.nc_average_sequence_length	Average non-coding sequence length	Mean length of all sequences	bp
-43	genebuild.nc_longest_gene_length	Longest non-coding gene	Length of longest non-coding gene	bp
-44	genebuild.nc_long_non_coding_genes	Long non-coding genes	Long genes not coding for proteins	integer
-45	genebuild.nc_misc_non_coding_genes	Misc. non-coding genes	Miscellaneous non-coding genes	integer
-46	genebuild.nc_non_coding_genes	Non-coding genes	Genes that don't code for proteins	integer
-47	genebuild.nc_shortest_gene_length	Shortest non-coding gene	Length of shortest gene	bp
-48	genebuild.nc_small_non_coding_genes	Small non-coding genes	Small genes not coding for proteins	integer
-49	genebuild.nc_total_introns	Introns in non-coding genes	Total intron count	integer
-50	genebuild.nc_total_transcripts	Non-coding transcripts	Total RNA transcript count	integer
-51	genebuild.nc_transcripts_per_gene	Average transcripts per non-coding gene	Mean transcripts count per gene	float
-52	genebuild.ps_average_exons_per_transcript	Average intron length per pseudogene	Mean exon count per pseudogene transcript	float
-53	genebuild.ps_average_exon_length	Average exon length per pseudogene	Mean pseudogene exon length	bp
-54	genebuild.ps_average_genomic_span	Average pseudogene genomic span	Mean length of pseudogene regions	bp
-55	genebuild.ps_average_intron_length	Average intron length per pseudogene	Mean pseudogene intron length	bp
-56	genebuild.ps_average_sequence_length	Average pseudogene sequence length	Mean length of pseudogene sequences	bp
-57	genebuild.ps_longest_gene_length	Longest pseudogene	Length of longest pseudogene	bp
-58	genebuild.ps_pseudogenes	Pseudogenes	Genes which don't code functional protiens	integer
-59	genebuild.ps_shortest_gene_length	Shortest pseudogene	Length of shortest pseudogene	bp
-60	genebuild.ps_total_exons	Exons in pseudogenes	Total exon count in pseudogenes	integer
-61	genebuild.ps_total_introns	Introns in pseudogenes	Total intron count in pseudogenes	integer
-62	genebuild.ps_total_transcripts	Transcripts in pseudogenes	Total pseudogene RNA transcript count	integer
-63	genebuild.ps_transcripts_per_gene	Average transcripts per pseudogene	Mean pseudogene transcripts count per pseudogene	float
-64	genebuild.shortest_gene_length	Shortest coding gene	Length of shortest gene	bp
+39	genebuild.stats.nc_average_exons_per_transcript	Average exons per non-coding transcript	Mean exon count per transcript	float
+40	genebuild.stats.nc_average_exon_length	Average exon length per non-coding transcript	Mean exon length	bp
+41	genebuild.stats.nc_average_genomic_span	Average non-coding genomic span	Mean length of all genomic regions	bp
+42	genebuild.stats.nc_average_sequence_length	Average non-coding sequence length	Mean length of all sequences	bp
+43	genebuild.stats.nc_longest_gene_length	Longest non-coding gene	Length of longest non-coding gene	bp
+44	genebuild.stats.nc_long_non_coding_genes	Long non-coding genes	Long genes not coding for proteins	integer
+45	genebuild.stats.nc_misc_non_coding_genes	Misc. non-coding genes	Miscellaneous non-coding genes	integer
+46	genebuild.stats.nc_non_coding_genes	Non-coding genes	Genes that don't code for proteins	integer
+47	genebuild.stats.nc_shortest_gene_length	Shortest non-coding gene	Length of shortest gene	bp
+48	genebuild.stats.nc_small_non_coding_genes	Small non-coding genes	Small genes not coding for proteins	integer
+49	genebuild.stats.nc_total_introns	Introns in non-coding genes	Total intron count	integer
+50	genebuild.stats.nc_total_transcripts	Non-coding transcripts	Total RNA transcript count	integer
+51	genebuild.stats.nc_transcripts_per_gene	Average transcripts per non-coding gene	Mean transcripts count per gene	float
+52	genebuild.stats.ps_average_exons_per_transcript	Average intron length per pseudogene	Mean exon count per pseudogene transcript	float
+53	genebuild.stats.ps_average_exon_length	Average exon length per pseudogene	Mean pseudogene exon length	bp
+54	genebuild.stats.ps_average_genomic_span	Average pseudogene genomic span	Mean length of pseudogene regions	bp
+55	genebuild.stats.ps_average_intron_length	Average intron length per pseudogene	Mean pseudogene intron length	bp
+56	genebuild.stats.ps_average_sequence_length	Average pseudogene sequence length	Mean length of pseudogene sequences	bp
+57	genebuild.stats.ps_longest_gene_length	Longest pseudogene	Length of longest pseudogene	bp
+58	genebuild.stats.ps_pseudogenes	Pseudogenes	Genes which don't code functional protiens	integer
+59	genebuild.stats.ps_shortest_gene_length	Shortest pseudogene	Length of shortest pseudogene	bp
+60	genebuild.stats.ps_total_exons	Exons in pseudogenes	Total exon count in pseudogenes	integer
+61	genebuild.stats.ps_total_introns	Introns in pseudogenes	Total intron count in pseudogenes	integer
+62	genebuild.stats.ps_total_transcripts	Transcripts in pseudogenes	Total pseudogene RNA transcript count	integer
+63	genebuild.stats.ps_transcripts_per_gene	Average transcripts per pseudogene	Mean pseudogene transcripts count per pseudogene	float
+64	genebuild.stats.shortest_gene_length	Shortest coding gene	Length of shortest gene	bp
 65	genebuild.start_date	genebuild.start_date	genebuild.start_date	string
-66	genebuild.total_coding_exons	Exons in coding genes	Total number of coding exons	integer
-67	genebuild.total_exons	Exons in genes	Total number of exons	integer
-68	genebuild.total_introns	Introns in coding genes	Total number of introns	integer
-69	genebuild.total_transcripts	Transcripts in coding genes	Total number of RNA transcripts	integer
-70	genebuild.transcripts_per_gene	Average transcripts per coding gene	Average number of transcripts per gene	float
+66	genebuild.stats.total_coding_exons	Exons in coding genes	Total number of coding exons	integer
+67	genebuild.stats.total_exons	Exons in genes	Total number of exons	integer
+68	genebuild.stats.total_introns	Introns in coding genes	Total number of introns	integer
+69	genebuild.stats.total_transcripts	Transcripts in coding genes	Total number of RNA transcripts	integer
+70	genebuild.stats.transcripts_per_gene	Average transcripts per coding gene	Average number of transcripts per gene	float
 71	genebuild.version	genebuild.version	genebuild.version	string
 72	genebuild.sample_gene	genebuild.sample_gene	Sample Gene Data	string
 73	genebuild.sample_location	genebuild.sample_location	Sample Location Data	string
 74	assembly.coverage_depth	assembly.coverage_depth	assembly.coverage_depth	string
 75	assembly.web_accession_source	assembly.web_accession_source	assembly.web_accession_source	string
 76	assembly.web_accession_type	assembly.web_accession_type	assembly.web_accession_type	string
 77	genebuild.id	genebuild.id	genebuild.id	string
-78	genebuild.nc_average_intron_length	Average intron length per non-coding transcript	Mean intron length	bp
+78	genebuild.stats.nc_average_intron_length	Average intron length per non-coding transcript	Mean intron length	bp
 79	genebuild.projection_source_db	genebuild.projection_source_db	genebuild.projection_source_db	string
 80	assembly.long_name	assembly.long_name	assembly.long_name	string
 81	assembly.url_name	assembly.url_name	assembly.url_name	string
 82	genebuild.havana_datafreeze_date	genebuild.havana_datafreeze_date	genebuild.havana_datafreeze_date	string
 83	assembly.version	assembly.version	assembly.version	string
 84	genebuild.provider_name	genebuild.provider_name	genebuild.provider_name	string
 85	genebuild.provider_url	genebuild.provider_url	genebuild.provider_url	string
-119	variation.short_variants	Short variants	Small-scale genetic variations	integer
+119	variation.stats.short_variants	Short variants	Small-scale genetic variations	integer
 120	variation.sample_variant	variation.sample_variant	variation.sample_variant	string
-123	variation.short_variants_with_phenotype_assertions	Short variation with phenotype assertion	Short variation with phenotype assertion	string
-161	compara.homology_coverage	compara.homology_coverage	compara.homology_coverage	float
+123	variation.stats.short_variants_with_phenotype_assertions	Short variation with phenotype assertion	Short variation with phenotype assertion	string
+161	compara.stats.homology_coverage	compara.stats.homology_coverage	compara.stats.homology_coverage	float
 162	compara.homology_reference_species	compara.homology_reference_species	compara.homology_reference_species	string
 163	regulation.open_chromatin_count	regulation.open_chromatin_count	Number of open chromatin regions	integer
-164	regulation.promoter_count	regulation.promoter_count	Number of promoters	integer
-165	regulation.enhancer_count	regulation.enhancer_count	Number of enhancers	integer
-166	regulation.ctcf_count	regulation.ctcf_count	Number of CTCF binding sites	integer
+164	regulation.stats.promoter_count	regulation.stats.promoter_count	Number of promoters	integer
+165	regulation.stats.enhancer_count	regulation.stats.enhancer_count	Number of enhancers	integer
+166	regulation.stats.ctcf_count	regulation.stats.ctcf_count	Number of CTCF binding sites	integer
 167	regulation.tfbs_count	regulation.tfbs_count	Number of regions enriched for transcription factor binding	integer
 168	assembly.tol_id	assembly.tol_id	assembly.tol_id	string
 169	genebuild.annotation_source	genebuild.annotation_source	genebuild.annotation_source	string
-170	genebuild.nc_total_exons	Exons in non-coding genes	Total exon count	integer
+170	genebuild.stats.nc_total_exons	Exons in non-coding genes	Total exon count	integer
 179	assembly.description	assembly.description	assembly.description	string
 180	assembly.master_accession	assembly.master_accession	assembly.master_accession	string
 181	assembly.alt_accession	assembly.alt_accession	assembly.alt_accession	string