TogoMCP

schema_info: title: UniProt RDF description: 'Comprehensive protein sequence and functional information integrating Swiss-Prot (manually curated) and TrEMBL (automatically annotated). Contains 444M proteins with sequences, functions, domains, structures, variants, and cross-references to 200+ databases. CRITICAL: reviewed=1 (Swiss-Prot, 923K entries) vs reviewed=0 (TrEMBL, 444M entries). Always filter by reviewed=1 for quality and performance. ' endpoint: https://rdfportal.org/backend/sib/sparql base_uri: http://purl.uniprot.org/ graphs: - http://sparql.uniprot.org/uniprot version: mie_version: '1.2' mie_created: '2024-12-08' data_version: Release 2024_06 update_frequency: Monthly license: data_license: Creative Commons Attribution 4.0 International license_url: https://www.uniprot.org/help/license access: rate_limiting: Reasonable use policy max_query_timeout: 60 seconds backend: Virtuoso (supports bif:contains) shape_expressions: | PREFIX up: <http://purl.uniprot.org/core/> <ProteinShape> { a [ up:Protein ] ; dcterms:identifier xsd:string ; up:mnemonic xsd:string ; up:organism @<TaxonShape> ; up:sequence @<IsoformShape> + ; up:annotation @<AnnotationShape> * ; up:classifiedWith IRI * ; up:encodedBy @<GeneShape> ? ; up:reviewed xsd:integer ; up:recommendedName @<NameShape> ? ; up:enzyme IRI * ; rdfs:seeAlso IRI * } <IsoformShape> { a [ up:Simple_Sequence ] ; rdf:value xsd:string ; up:mass xsd:integer ; up:md5Checksum xsd:string } <NameShape> { a [ up:Structured_Name ] ; up:fullName xsd:string ; up:shortName xsd:string ? } <TaxonShape> { a [ up:Taxon ] ; up:scientificName xsd:string ; rdfs:subClassOf IRI * } <AnnotationShape> { a [ up:Annotation ] ; rdfs:comment xsd:string ? } <GeneShape> { a [ up:Gene ] ; skos:prefLabel xsd:string ? } sample_rdf_entries: - title: Reviewed Human P53 Protein description: Swiss-Prot tumor suppressor with expert curation and PDB structures. rdf: | uniprot:P04637 a up:Protein ; dcterms:identifier "P04637" ; up:mnemonic "P53_HUMAN" ; up:organism taxon:9606 ; up:reviewed 1 ; up:version 307 ; rdfs:seeAlso <http://rdf.wwpdb.org/pdb/1H26> . - title: Protein Sequence description: Canonical isoform with validated amino acid sequence. rdf: | isoforms:P04637-1 a up:Simple_Sequence ; rdf:value "MEEPQSDPSVEPPLSQETFSDLWKLLPENNVLSPLPSQAMDDLMLSPDDIEQWF..." ; up:mass 43653 ; up:md5Checksum "c133dfce69f606f20865e9008199f852" . - title: Function Annotation description: Expert-curated functional annotation from Swiss-Prot. rdf: | annotation:12345 a up:Function_Annotation ; rdfs:comment "Acts as a tumor suppressor; induces growth arrest or apoptosis..." . - title: Reviewed Enzyme description: Mitochondrial enzyme with Swiss-Prot quality assurance. rdf: | uniprot:P86925 a up:Protein ; up:mnemonic "RLGM2_TRYB2" ; up:organism taxon:185431 ; up:reviewed 1 ; up:recommendedName [ up:fullName "RNA-editing ligase 2, mitochondrial" ] . - title: Kinase with Structures description: Swiss-Prot kinase with multiple validated PDB structures. rdf: | uniprot:P17612 a up:Protein ; up:mnemonic "KAPCA_HUMAN" ; up:reviewed 1 ; rdfs:seeAlso <http://rdf.wwpdb.org/pdb/6BYR> ; up:recommendedName [ up:fullName "cAMP-dependent protein kinase catalytic subunit alpha" ] . sparql_query_examples: - title: Find Reviewed Proteins by Organism description: Retrieve expert-curated human proteins from Swiss-Prot. question: What high-quality human proteins exist? complexity: basic sparql: | PREFIX up: <http://purl.uniprot.org/core/> SELECT ?protein ?mnemonic WHERE { ?protein a up:Protein ; up:mnemonic ?mnemonic ; up:reviewed 1 ; up:organism <http://purl.uniprot.org/taxonomy/9606> . } LIMIT 30 - title: Search by Function with bif:contains description: Find reviewed proteins by validated functional descriptions using full-text search. Note that bif:contains requires splitting property paths. question: Which reviewed proteins are kinases or involved in DNA repair? complexity: basic sparql: | PREFIX up: <http://purl.uniprot.org/core/> SELECT DISTINCT ?protein ?mnemonic ?fullName WHERE { ?protein a up:Protein ; up:mnemonic ?mnemonic ; up:reviewed 1 ; up:recommendedName ?name . ?name up:fullName ?fullName . ?fullName bif:contains "'kinase' OR 'dna repair'" } LIMIT 15 - title: Get Functional Annotations description: Retrieve comprehensive functional annotations and GO terms for specific proteins. question: What are functions and GO terms for P04637, P17612, P86925? complexity: intermediate sparql: | PREFIX up: <http://purl.uniprot.org/core/> PREFIX uniprot: <http://purl.uniprot.org/uniprot/> SELECT ?protein ?mnemonic ?functionComment ?goLabel WHERE { VALUES ?protein { uniprot:P04637 uniprot:P17612 uniprot:P86925 } ?protein up:mnemonic ?mnemonic . OPTIONAL { ?protein up:annotation ?annot . ?annot a up:Function_Annotation ; rdfs:comment ?functionComment . } OPTIONAL { ?protein up:classifiedWith ?goTerm . ?goTerm rdfs:label ?goLabel . FILTER(STRSTARTS(STR(?goTerm), "http://purl.obolibrary.org/obo/GO_")) } } LIMIT 20 - title: Get Protein Sequences description: Retrieve expert-validated amino acid sequences with molecular properties. question: What are verified sequences and characteristics? complexity: intermediate sparql: | PREFIX up: <http://purl.uniprot.org/core/> SELECT DISTINCT ?protein ?mnemonic ?sequence ?mass ?checksum WHERE { ?protein a up:Protein ; up:mnemonic ?mnemonic ; up:reviewed 1 ; up:sequence ?iso . ?iso rdf:value ?sequence ; up:mass ?mass ; up:md5Checksum ?checksum . } LIMIT 10 - title: Count Tumor Suppressors with bif:contains description: Quantify expert-curated human tumor suppressors using full-text search. Note that bif:contains requires splitting property paths. question: How many human tumor suppressors in Swiss-Prot? complexity: intermediate sparql: | PREFIX up: <http://purl.uniprot.org/core/> SELECT (COUNT(*) as ?count) WHERE { ?protein a up:Protein ; up:reviewed 1 ; up:organism <http://purl.uniprot.org/taxonomy/9606> ; up:annotation ?annot . ?annot rdfs:comment ?function . ?function bif:contains "'tumor suppressor'" } - title: Enzyme Classification with GO description: Link enzymatic classifications with Gene Ontology cellular components. question: Which enzymes have specific GO localizations? complexity: advanced sparql: | PREFIX up: <http://purl.uniprot.org/core/> SELECT DISTINCT ?protein ?mnemonic ?fullName ?enzyme ?goLabel WHERE { ?protein a up:Protein ; up:mnemonic ?mnemonic ; up:reviewed 1 ; up:enzyme ?enzyme ; up:recommendedName/up:fullName ?fullName ; up:classifiedWith ?goTerm . ?goTerm rdfs:label ?goLabel . FILTER(STRSTARTS(STR(?goTerm), "http://purl.obolibrary.org/obo/GO_")) } LIMIT 20 - title: Taxonomic Hierarchy description: Navigate protein taxonomic relationships showing organism lineage. question: What is taxonomic classification hierarchy for human proteins? complexity: advanced sparql: | PREFIX up: <http://purl.uniprot.org/core/> SELECT DISTINCT ?protein ?mnemonic ?organism ?scientificName ?parentName WHERE { ?protein a up:Protein ; up:mnemonic ?mnemonic ; up:reviewed 1 ; up:organism ?organism . ?organism up:scientificName ?scientificName ; rdfs:subClassOf/up:scientificName ?parentName . FILTER(?organism = <http://purl.uniprot.org/taxonomy/9606>) } LIMIT 10 cross_references: - pattern: rdfs:seeAlso description: Links to 200+ external databases via rdfs:seeAlso. Filter by URL substring for specific databases. databases: structure: - PDB (rdf.wwpdb.org, ~14-25%) - AlphaFold (alphafolddb, >98%) sequence: - EMBL (~95%) - RefSeq (~80%) - Ensembl (high) genes: - HGNC (100% human) - neXtProt (100% human) - Gene IDs (~90%) families: - InterPro (>98%) - Pfam (~85%) - PANTHER (~80%) interactions: - IntAct (~16%) - STRING (~85%) - BioGRID (~25%) pathways: - KEGG (~95%) - Reactome (~30%) sparql: | SELECT ?protein ?mnemonic ?externalDB WHERE { ?protein a up:Protein ; up:mnemonic ?mnemonic ; up:reviewed 1 ; rdfs:seeAlso ?externalDB . # Examples: FILTER(CONTAINS(STR(?externalDB), "rdf.wwpdb.org")) } LIMIT 30 - pattern: up:organism description: Links proteins to NCBI Taxonomy classification. All proteins have exactly one organism. databases: taxonomy: - NCBI Taxonomy (100%) sparql: | SELECT ?protein ?taxon ?scientificName WHERE { ?protein a up:Protein ; up:reviewed 1 ; up:organism ?taxon . ?taxon up:scientificName ?scientificName . } LIMIT 30 - pattern: up:classifiedWith description: Ontology-based classifications, primarily Gene Ontology (GO) terms. databases: ontology: - Gene Ontology (>85% reviewed) - Other classifications sparql: | SELECT ?protein ?classification ?label WHERE { ?protein a up:Protein ; up:reviewed 1 ; up:classifiedWith ?classification . ?classification rdfs:label ?label . FILTER(STRSTARTS(STR(?classification), "http://purl.obolibrary.org/obo/GO_")) } LIMIT 30 architectural_notes: schema_design: - Central up:Protein entity with reviewed property as quality indicator (0=TrEMBL, 1=Swiss-Prot) - Hierarchical taxonomy via rdfs:subClassOf relationships - Annotation system with specialized subtypes - Cross-references via rdfs:seeAlso to 200+ databases performance: - 'CRITICAL: Always filter by up:reviewed 1 (reduces 444M to 923K, 99.8% reduction)' - COUNT queries require reviewed=1 filter to prevent timeout - Use bif:contains for keyword searches but split property paths (do not use /) - Organism filtering optimized with reviewed status data_integration: - Swiss-Prot has vastly superior cross-reference coverage - GO integration most reliable for reviewed proteins (>85% vs ~30%) - PDB structures predominantly for reviewed entries (~14-25% vs <1%) data_quality: - 'reviewed=1: expert manual curation vs automated TrEMBL' - Functional annotations 90%+ complete for Swiss-Prot vs 20-30% TrEMBL - Use up:organism for organism filtering, never mnemonic suffixes data_statistics: total_proteins: 444565015 reviewed_proteins: 923147 human_reviewed: 40209 coverage: reviewed_with_sequences: '>99%' reviewed_with_function: '>90%' reviewed_with_go: '>85%' reviewed_with_pdb: ~14-25% reviewed_with_alphafold: '>98%' reviewed_with_interpro: '>98%' cardinality: avg_isoforms_per_protein: 1.1 avg_pdb_per_protein: 4.8 avg_go_terms_per_protein: 12.5 avg_xrefs_per_protein: 45 performance_characteristics: - COUNT requires reviewed=1 filter - Reviewed filtering reduces dataset 99.8% - bif:contains for efficient text search (requires split property paths) anti_patterns: - title: Querying Without Reviewed Filter problem: Queries without up:reviewed 1 timeout and return unreliable automated annotations. wrong_sparql: | SELECT (COUNT(*) as ?count) WHERE { ?protein a up:Protein ; up:organism <http://purl.uniprot.org/taxonomy/9606> . } correct_sparql: | SELECT (COUNT(*) as ?count) WHERE { ?protein a up:Protein ; up:reviewed 1 ; up:organism <http://purl.uniprot.org/taxonomy/9606> . } explanation: Adding up:reviewed 1 reduces from 444M to 923K entries (99.8% reduction), enabling COUNT and ensuring expert-curated quality. - title: Using bif:contains with Property Paths problem: bif:contains does not work with SPARQL property paths (/) and causes 400 errors. wrong_sparql: | SELECT ?protein ?fullName WHERE { ?protein up:reviewed 1 ; up:recommendedName/up:fullName ?fullName . ?fullName bif:contains "'kinase'" } correct_sparql: | SELECT ?protein ?fullName WHERE { ?protein up:reviewed 1 ; up:recommendedName ?name . ?name up:fullName ?fullName . ?fullName bif:contains "'kinase'" } explanation: Split property paths into separate triple patterns when using bif:contains. The magic predicate requires variables directly bound to RDF literals. - title: Using Mnemonic for Organism Filtering problem: Filtering organisms via mnemonic suffixes is unreliable and incomplete. wrong_sparql: | SELECT ?protein ?mnemonic WHERE { ?protein up:reviewed 1 ; up:mnemonic ?mnemonic . FILTER(CONTAINS(?mnemonic, "_HUMAN")) } correct_sparql: | SELECT ?protein ?mnemonic WHERE { ?protein up:reviewed 1 ; up:organism <http://purl.uniprot.org/taxonomy/9606> ; up:mnemonic ?mnemonic . } explanation: Use up:organism with exact taxonomy URIs. Mnemonics are display-only, not semantic identifiers. common_errors: - error: Query timeout causes: - Missing up:reviewed 1 filter - COUNT/aggregation on full 444M dataset - Using CONTAINS() instead of bif:contains for large text searches solutions: - Always add up:reviewed 1 as first filter - Use bif:contains for keyword filtering (but split property paths) - Combine reviewed with organism filter - Use LIMIT (30-50 recommended) - error: bif:contains returns 400 Bad Request causes: - Using bif:contains with property paths (e.g., up:recommendedName/up:fullName) - Property path creates intermediate bindings that bif:contains cannot handle solutions: - Split property paths into separate triple patterns - Example WRONG - up:annotation/rdfs:comment ?text . ?text bif:contains 'keyword' - Example CORRECT - up:annotation ?a . ?a rdfs:comment ?text . ?text bif:contains 'keyword' - error: Empty organism results causes: - Using mnemonic pattern matching instead of up:organism - Incorrect taxonomy URI format solutions: - Use up:organism <http://purl.uniprot.org/taxonomy/TAXID> - Never filter by mnemonic text patterns - Always include up:reviewed 1 - error: Missing GO terms causes: - Incorrect GO URI FILTER syntax - Not using OPTIONAL for GO terms - Missing reviewed filter (lower GO coverage) solutions: - Use STRSTARTS(STR(?goTerm), 'http://purl.obolibrary.org/obo/GO_') - Wrap GO patterns in OPTIONAL if not required - Include up:reviewed 1 for >85% GO coverage