Towards FAIRer Biological Knowledge Networks
Using a Hybrid Linked Data
and Graph Database approach
Harpenden, 5/6/2018
Marco Brandizi <marco.brandizi@rothamsted.ac.uk>
Find these slides on SlideShare
KnetMiner-inspired Artwork
by Hugo Dalton (hugodalton.com)

Can we do More with KnetMiner Data?
(and better)

Behind the Scenes
• Starting point: graph data model
• With concepts, relations between concepts hierarchies of concept classes and relation
types
• => There are standardised ways for it
• Make app development easier
• independent components on top of a unified data model
• clear separation between data access and apps
• Serve third-party applications, making their data access no different than ours
• Simplify the way we ingest data,
• ease conversions from multiple formats into unified model
• relax the high-memory requirements need (e.g., backing data store)
• prepare for scalability (e.g., cloud stores, big data stores)

Putting it on a Bigger Picture

The Semantic Web Way
• It’s for networked knowledge (semantic networks)
• Focuses on sharing via web technologies and
principles (eg, share resolvable URIs)
• Rich ‘schema’ language, already much used in life
sciences (i.e., ontologies, coming from frames and 1st-
order logics)
• protocol + a standard query language (SPARQL)

Modelling data with OWL:
Promises & Wishes

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared
• Fairly reasonable: all those vocabularies are headaches, I don’t have expertise

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared
• Fairly reasonable: all those vocabularies are headaches, I don’t have expertise
• Reasonable: I have a different point of view

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared
• Fairly reasonable: all those vocabularies are headaches, I don’t have expertise
• Reasonable: I have a different point of view
• Not always reasonable: possibly, yours is complicated/wrong/stupid/idiotic/
worse

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared
• Fairly reasonable: all those vocabularies are headaches, I don’t have expertise
• Reasonable: I have a different point of view
• Not always reasonable: possibly, yours is complicated/wrong/stupid/idiotic/
worse
• Likely not reasonable: Not Invented Here

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared
• Fairly reasonable: all those vocabularies are headaches, I don’t have expertise
• Reasonable: I have a different point of view
• Not always reasonable: possibly, yours is complicated/wrong/stupid/idiotic/
worse
• Likely not reasonable: Not Invented Here
• No comment: if I reinvent it, I can publish it

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared
• Fairly reasonable: all those vocabularies are headaches, I don’t have expertise
• Reasonable: I have a different point of view
• Not always reasonable: possibly, yours is complicated/wrong/stupid/idiotic/
worse
• Likely not reasonable: Not Invented Here
• No comment: if I reinvent it, I can publish it
• Just joking (maybe…): Your ontology is good, but I’d rather stab you on your back

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared
• Fairly reasonable: all those vocabularies are headaches, I don’t have expertise
• Reasonable: I have a different point of view
• Not always reasonable: possibly, yours is complicated/wrong/stupid/idiotic/
worse
• Likely not reasonable: Not Invented Here
• No comment: if I reinvent it, I can publish it
• Just joking (maybe…): Your ontology is good, but I’d rather stab you on your back
In fact, they did this

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared
• Fairly reasonable: all those vocabularies are headaches, I don’t have expertise
• Reasonable: I have a different point of view
• Not always reasonable: possibly, yours is complicated/wrong/stupid/idiotic/
worse
• Likely not reasonable: Not Invented Here
• No comment: if I reinvent it, I can publish it
• Just joking (maybe…): Your ontology is good, but I’d rather stab you on your back
In fact, they did this

Modelling data with OWL:
Promises & Wishes
• Rich semantics, but also very formal, so that powerful automated reasoning is possible
• On the messy web ocean?!
• What about performance?!
• Very formal semantics is not very easy:
• SomeValuesFrom Restriction?!
• My blood sample derives from some Skolem human, not from
NCBI:HomoSapiens?
• Ontologies defined for the whole world and then harmoniously and lovely shared
• Fairly reasonable: all those vocabularies are headaches, I don’t have expertise
• Reasonable: I have a different point of view
• Not always reasonable: possibly, yours is complicated/wrong/stupid/idiotic/
worse
• Likely not reasonable: Not Invented Here
• No comment: if I reinvent it, I can publish it
• Just joking (maybe…): Your ontology is good, but I’d rather stab you on your back
In fact, they did this

Simplifying Views in BioKNO
obo:GO_0030015
a owl:Class ;
rdfs:label "CCR4-NOT core complex"^^xsd:string ;
rdfs:subClassOf obo:GO_0044424, obo:GO_0044424,
[
a owl:Restriction ;
owl:onProperty <http://purl.obolibrary.org/obo/BFO_0000050> ; # 'part of'
owl:someValuesFrom obo:GO_0030014 # CCR4-NOT complex
] ;
oboInOwl:id "GO:0030015"^^xsd:string ;
obo:IAO_0000115 "The core of the CCR4-NOT complex. In Saccharomyces the CCR4-NOT...";
oboInOwl:hasOBONamespace "cellular_component"^^xsd:string .

Simplifying Views in BioKNO
obo:GO_0030015
a owl:Class ;
rdfs:label "CCR4-NOT core complex"^^xsd:string ;
rdfs:subClassOf obo:GO_0044424, obo:GO_0044424,
[
a owl:Restriction ;
owl:onProperty <http://purl.obolibrary.org/obo/BFO_0000050> ; # 'part of'
owl:someValuesFrom obo:GO_0030014 # CCR4-NOT complex
] ;
oboInOwl:id "GO:0030015"^^xsd:string ;
obo:IAO_0000115 "The core of the CCR4-NOT complex. In Saccharomyces the CCR4-NOT...";
oboInOwl:hasOBONamespace "cellular_component"^^xsd:string .
obo:GO_0030014 a bk:GeneOntologyTerm ;
dc:identifier obo:GO_0030014_acc ;
bk:is_a obo:GO_0044424 , obo:GO_0043234 ;
bk:prefName "CCR4-NOT complex" .
obo:GO_0030015 a bk:GeneOntologyTerm;
bk:prefName "CCR4-NOT core complex";
bk:is_a obo:GO_0044424, obo:GO_0043234 ;
bk:part_of obo:GO_0030014;
dc:identifier obo:GO_0030015_acc.
obo:GO_0044424 a bk:GeneOntologyTerm;
bk:prefName "intracellular part" ;
• OWL is simplified mixing classes with SKOS-style
concepts
• More suitable for less formal, more simple
taxonomies
• OWL-2 punning makes it consistent

The BioKNO Ontology
(and The rest of the World)
BioKNO External Ontologies Mapping Type
bk:Concept skos:Concept Subclass
bk:Relation
bk:relFrom
bk:relTypeRef
bk:relTo
rdf:Statement
rdf:subject
rdf:predicate
rdf:object
Subclass
Subproperties
(ie, mapping to RDF reified
statements)
bk:Path, bk:Participant, bk:Interaction, bk:Transport,
bk:Protein, bk:Gene
Classes with same names in BioPAX and SIO Equivalent Class
bk:participates_in
bk:has_participant
Relation Ontology (RO) properties with same names
biopax:participant (as sub-property)
Equivalent property
bk:produces
bk:produced_by
bk:consumes
bk:consumed_by
biopax:product (as sub-property)
RO properties with same names
Equivalent property
bk:regulates
bk:positively_regulates
bk:negatively_regulates
RO properties with same names Equivalent property
bk:is_a
bk:part_of, bk:has_part
bk:occurs_in, bk:co_occurs_with
skos:broader
Basic Formal Ontology (BFO)/RO properties with same
names
Equivalent property
bk:Publication schema:CreativeWork Subclass
bka:abstract
bka:title (also known as AbstractHeader)
bka:authors
dcterms:description
dcterms:title
dc:creator
Sub-property

The BioKNO Ontology

Putting it on a Bigger Picture

Putting it on a Bigger Picture

Accessing RDF through SPARQL

Accessing RDF through SPARQL

Accessing RDF through SPARQL

CONSTRUCT {
?protIri bk:expressed_by ?sampleIri.
?degRelIri
a bk:Relation;
bka:PVALUE ?pValue;
bk:evidence bkev:EXP; # Inferred from experiment
bk:relFrom ?protIri; # Details defined by UniProt info
bk:relTo ?sampleIri; # Details defined by sample_degs_2.tsv
bk:relTypeRef bk:expressed_by.
}
WHERE {
# Some IDs and IRIs to be defined above
BIND ( LCASE ( REPLACE ( ?Sample, ' ', '_' ) ) AS ?sampleId )
BIND ( IRI ( CONCAT ( STR ( bkr: ), ?Gene_Symbol ) ) AS ?protIri )
BIND ( IRI ( CONCAT ( STR ( bkr: ), 'degex_', ?sampleId ) ) AS ?sampleIri )
BIND ( IRI ( CONCAT ( STR ( bkr: ), 'degex_', ?sampleId, '_', LCASE ( ?Gene_Symbol ) ) )
AS ?degRelIri )
BIND ( xsd:double ( ?p_value ) AS ?pValue )
}
Extraction, Loading, Transformation
SPARQL/TARQL Example

SPARQL/RDF for ELT
• RDF-to-RDF translation via CONSTRUCT (or SPARUL)
• TARQL: Using SPARQL to RDF-Convert Tabular CSV Files
• RDF/XML can be transformed via XSL
• We have done it for bio-specific ontology definitions in Ondex
• Programmatic conversions
• Using RDF frameworks, eg, Jena, RDF4J (former Sesame), rdflib for
Python
• See also java2rdf (https://github.com/EBIBioSamples/java2rdf)
• We have used it for the Ondex->RDF converter

SPARQL/RDF for ELT
• RDF-to-RDF translation via CONSTRUCT (or SPARUL)
• TARQL: Using SPARQL to RDF-Convert Tabular CSV Files
• RDF/XML can be transformed via XSL
• We have done it for bio-specific ontology definitions in Ondex
• Programmatic conversions
• Using RDF frameworks, eg, Jena, RDF4J (former Sesame), rdflib for
Python
• See also java2rdf (https://github.com/EBIBioSamples/java2rdf)
• We have used it for the Ondex->RDF converter

Issues
https://lod-cloud.net/

Issues
https://lod-cloud.net/
• Still not so popular (especially in more commercial contexts)
• It’s (perceived as) difficult (in particular, SPARQL)
• Bad reputation
• Performance can still be an issue
• eg, optimising SPARQL can be hard
• Specific issues
• eg, I need contextualised/attribute-attached properties
• and I don’t fancy reified relations…

Another Graph Database World
Property Graphs

Neo4j on top Of RDF

Application to Semantic Motif Search

The rdf2neo Tool
https://github.com/Rothamsted/rdf2neo

Triple Stores vs Prop Graphs
Neo4j, Cypher DBs, Graph DBs Semantic Web/Triple Stores
Data xchg format
- No official one, just Cypher,
Support for GraphML, RDF
+/- Focus on backing applications
+ Focus on data sharing standards
Data model
+ Relations with properties
- Metadata/schemas/ontologies management
- Relations cannot have properties (reification
required)
+ Metadata/schemas/ontologies as first citizen
and standardised OWL
Performance + complex graph traversals + Comparable in most cases
Query Language
+ Cypher is easier (eg, compact, implicit elems)?
- Expressivity issues (unions)
- No standard QL (but efforts in progress, eg,
OpenCypher)
- SPARQL is Harder? (URIs, namespaces,
verbosity)
+ SPARQL More expressive
Standardisation,
openness
+/- (TinkerPop is open, Neo4j isn’t)
+ Commercial support
+ More alive and up-to date (e.g., support for
Hadoop, nice Neo4j browser, easy installation)
+ Natively open, many open implementations
- Instability and many short-lived prototypes
- Advancements seems to be slowing down
+ Some nice open and commercial browser
(LODEStar,
Scalability,
big data
+/- Commercial support to clustering/clouds for
Neo4j
+ Open support in TinkerPop
+ Load Balancing/Cluster solutions, Commercial
Cloud support (eg GraphDB)
+ SPARQL Over TinkerPop (via SAIL inteface)

Bridging to RDF: JSON-LD
…
"@id": "bkr:TOB1",
"@type": "bk:Protein",
"prefName": "TOB1 Human",
"dcterms:identifier": "TOB1",
"is_annotated_by": "obo:GO_0030014",
"participates_in": {
"@id": "http://www.wikipathways.org/id1",
"@type": "bk:Pathway",
"evidence": "bkev:IMPD",
"prefName":
“Bone Morphogenic Protein (BMP) Signalling and Regulation"
}
}
{
"@context": {
"bk": "http://www.ondex.org/bioknet/terms/",
"bka": "http://www.ondex.org/bioknet/terms/attributes/",
"bkds": "http://www.ondex.org/bioknet/terms/dataSources/",
"bkev": "http://www.ondex.org/bioknet/terms/evidences/",
"bkr": "http://www.ondex.org/bioknet/resources/",
"dcterms": "http://purl.org/dc/terms/",
"obo": "http://purl.obolibrary.org/obo/",
"xsd": "http://www.w3.org/2001/XMLSchema#",
"@vocab": "http://www.ondex.org/bioknet/terms/",
"dcterms:identifier": { "@type": "xsd:string" },
"evidence": { "@type": “@id" }
},
…

KnetMiner UI Overview
Search Select Explore
Addressing FAIR
• Findable (or, Semantic Web is still useful)
• SPARQL endpoint
• which powers URI Resolution
• Dataset-level metadata (e.g., VoID)
• Mapping to Standard Ontologies
• Interested in contributing to existing standards
(e.g., Bioschemas)
• API/JSON-Schema formalisation
• Accessible
• Multiple access means (SPARQL, URIs, JSON APIs, Cypher)
• Triple Stores and Property Graphs are complementary, not
alternative
• Data dumps
• Interoperable (or, Sem Web is still useful)
• Unified model encoded in at least one common syntax (RDF)
• URIs are reused
• Mappings to ontologies
• Reusable
• All of the above, plus multiple interfaces under unified model
• Support to common graph languages (e.g., Cytoscape.js)
• Converters (e.g., our RDF conversion scripts/tools, rdf2neo)
• Open Data licences