A Prototype Knowledge Base for the Life Sciences

W3C Interest Group Note 4 June 2008

This version:
Latest version:
Previous version:
M. Scott Marshall, University of Amsterdam <marshall@science.uva.nl>
Eric Prud'hommeaux, W3C <eric@w3.org>
Alan Ruttenberg, Science Commons <alanruttenberg@gmail.com>
Jonathan Rees, Science Commons <jar@creativecommons.org>
Susie Stephens, Lilly <Stephens_Susie_M@lilly.com>
Matthias Samwald, Yale Center for Medical Informatics; DERI Galway; Semantic Web Company <samwald@gmx.at>
Kei-Hoi Cheung, Yale Center for Medical Informatics <kei.cheung@yale.edu>


The prototype we describe is a biomedical knowledge base, constructed for a demonstration at Banff WWW2007 , that integrates 15 distinct data sources using currently available Semantic Web technologies such as the W3C standard Web Ontology Language [OWL] and Resource Description Framework [RDF]. This report outlines which resources were integrated, how the knowledge base was constructed using free and open source triple store technology, how it can be queried using the W3C Recommended RDF query language SPARQL [SPARQL], and what resources and inferences are involved in answering complex queries. While the utility of the knowledge base is illustrated by identifying a set of genes involved in Alzheimer's Disease, the approach described here can be applied to any use case that integrates data from multiple domains.

Status of This Document

This section describes the status of this document at the time of its publication. Other documents may supersede this document. A list of current W3C publications and the latest revision of this technical report can be found in the W3C technical reports index at http://www.w3.org/TR/.

This W3C Interest Group Note describes how one can use the Semantic Web to express and integrate scientific data. These techniques can be used for modeling any data, and the benefits of integration and model consistency apply to other diverse, distributed data domains. It is hoped that this document will inspire further contributions to the ongoing work at Neurocommons and the Health Care and Life Sciences Interest Group, as well as inspire those in other domains to exploit the Semantic Web.

This document describes the construction and use of the HCLS Knowledgebase used in the WWW2007 Banff HCLS Demo. It describes the process for creating a bilogical database on the Semantic Web. The companion document, Experiences with the conversion of SenseLab databases to RDF/OWL, describes the process for integrating new data into this Knowledgebase.

The document was produced by the Semantic Web in Health Care and Life Sciences Interest Group (HCLS), part of the W3C Semantic Web Activity (see charter). Comments may be sent to the publicly archived public-semweb-lifesci@w3.org mailing list. Feedback is encouraged, as is participation in the recently re-chartered HCLSIG. A list of changes since the last publication is available.

Publication as an Interest Group Note does not imply endorsement by the W3C Membership. This is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to cite this document as other than work in progress.

This document was produced by a group operating under the disclosure obligations of the 5 February 2004 W3C Patent Policy. The group does not expect this document to become a W3C Recommendation. An individual who has actual knowledge of a patent which the individual believes contains Essential Claim(s) must disclose the information to public-semweb-lifesci@w3.org [public archive] in accordance with in accordance with section 6 of the W3C Patent Policy.

Table of Contents


1 Introduction

The life sciences have a rich history of making data available on the Web, because researchers recognized the benefits of sharing data and made it available to other researchers for the benefit of greater science. However, because many of the data repositories were developed in relative isolation, they tend to use different identifier schemes, incompatible terminology, and dissimilar data formats. This makes it hard for researchers to find all data about an entity of interest and to assemble it into a useful block of knowledge. This prototype was built to demonstrate how Semantic Web technologies can integrate such heterogeneous data sets and thereby help scientists to more easily answer interesting scientific questions.

The key to advancing scientific understanding is empowering scientists with the information that they need to make well-informed decisions. Scientists need to be able to easily gain access to all information about chemical compounds, biological systems, diseases, and the interactions between these entities, and this requires data to be effectively integrated in order to provide a biological systems level view to the user, i.e. a complete view of biological activity. However, achieving this goal has proven to be a formidable challenge in the life sciences, where data and models are found in a large variety of formats and scales that span from the molecular to the anatomical.

In order to overcome the challenge of gaining insight directly from the Web, a number of laboratories, organizations, and companies have built internal data warehouses from the publicly available data sources. This certainly helps scientists to more easily query for all information related to entities of interest. However, these efforts generally integrate only a subset of publicly available data that is deemed to be of greatest interest, and it has proven difficult to add data sources to the warehouse at a later point. Further, advances in scientific knowledge require regular changes to be made to the underlying data models, and this is not straightforward with a relational model. Organizations that use this approach also typically face challenges with representing data that is at different levels of abstraction, and that includes data of very different quality.

Many health care and life sciences organizations are interested in the data integration abilities promised by the Semantic Web. More specifically, the benefits include the aggregation of heterogeneous data using explicit semantics, and the expression of rich and well-defined models for data aggregation and search. Semantic Web technologies enable one to more flexibly add additional data sets into the data model, and more easily reuse data in unanticipated ways. Once data has been aggregated, a Semantic Web reasoner computes implied relationships among the aggregated data resulting in tighter integration and the possibility of additional insights.

This prototype knowledge base imports data from data sources that span multiple domains in the life sciences to make cross-discipline queries. It therefore provides a working (and reproducible) example of the possibilities that become available via knowledge integration. The use of an RDF repository to store RDF and OWL makes it possible to query, manipulate, and reason about the data with standard tools, such as OWL reasoners, and languages, such as the SPARQL Query Language for RDF. Although this document addresses a specific use case, the approach described here can be applied to any use case that integrates data from multiple domains.

1.2 Document Scope and Target Audience

This document attempts to succinctly describe how this knowledge base was constructed so that interested parties can use the core techniques to create their own knowledge base. We have attempted to write a general description but, unavoidably, the knowledge base makes use of specialized resources, such as those found in the Data Sources section. Some, but not all, of the reasoning behind design decisions is explained. Several technologies such as the Semantic Web standards RDF, OWL, and SPARQL were used, but in order to keep this document to a manageable size, we will not explain all aspects in the depth that would be required for those new to the area. Those interested in a general introduction to the Semantic Web should see The Semantic Web Primer. See also the CO-ODE web site for a hands-on OWL tutorial with Protégé. For materials introducing ontology see National Center for Biomedical Ontology(NCBO) Introduction to Biomedical Ontologies. For materials related to reasoning see The Semantic Web: Ontologies and OWL.

1.3 Stability of Terms

This document uses URLs to identify records about biological entities and processes. The identifiers used in this document are the same as those used in the prototype knowledge base and are not yet stable. Knowledge base implementors should use these terms whenever possible.

1.4 Document Conventions

RDF data in this document is expressed in Turtle [TURTLE]. Queries on this data are expressed in SPARQL [SPARQL]. The following namespace prefix bindings are assumed unless otherwise stated:

Prefix URI Description
rdf: http://www.w3.org/1999/02/22-rdf-syntax-ns# The RDF Vocabulary
rdfs: http://www.w3.org/2000/01/rdf-schema# The RDF Schema vocabulary
xsd: http://www.w3.org/2001/XMLSchema# XML Schema
sc: http://purl.org/science/owl/sciencecommons/ The ad hoc Science Commons ontology
pubmedRec: http://purl.org/commons/record/pmid/ PubMed records (not the articles themselves)
article: http://purl.org/science/article/pmid/ PubMed articles
ncbi_gene: http://purl.org/commons/record/ncbi_gene/ Entrez Gene records (not the genes themselves)
proteinsubclass: http://purl.org/science/protein/subjects/ Proteins of a given gene participating in a given pathway
go: http://purl.org/obo/owl/GO# Gene Ontology terms
protein: http://purl.org/science/protein/bysequence/ National Center for Biotechnology Information (NCBI) records for Genes sequences
ro: http://www.obofoundry.org/ro/ro.owl# (proposed update may be more complete) Relation Ontology (RO): Relationships between members of OBO classes
obo: http://purl.org/obo/owl/obo# Open Biomedical Ontologies (OBO)
senselab: http://purl.org/ycmi/senselab/neuron_ontology.owl# Neuroscience ontology derived from the SenseLab NeuronDB database
dnaGeneProduct: http://purl.org/science/owl/sciencecommons/is_protein_gene_product_of_dna_ Syntactic trick to shorten sc:is_protein...described_by

1.5 Document Outline

1 Introduction motivates and explains this document.

2 Use Case introduces an interesting scientific question that the knowledge base can be used to address.

3 Data Sources describes the data sources that have been incorporated into the knowledge base.

4 Design Decisions explains the reasons for several design choices.

5 Importing to RDF - Homologene Example explains the process of translating data into RDF triples.

6 Query explains the use case query that answers the scientific question.

7 Data Model explains the basics of RDF triples.

8 Adding a New Data Source explains how the SenseLab database was integrated.

9 Named Graphs discusses the use of named graphs and query details.

10 Opportunities for further development discusses problem areas and possible improvements.

2 Use Case

Alzheimer's is a debilitating neurodegenerative disease that affects approximately 27 million people worldwide. The cause of Alzheimer's is currently unknown and no therapy is able to halt its progression. However, insight into the mechanism and potential treatment of this debilitating disease may come from the integration of neurological, biomedical and biological resources. The knowledge base assembles several neurology-related resources alongside an array of clinical and biological resources. This makes it possible to integrate knowledge across several research domains and potentially provide insight into the mechanisms of the disease.

The scientific question under scrutiny in our use case involves several elements of putative functional importance to Alzheimer's. CA1 Pyramidal Neurons (CA1PN) are known to be particularly damaged in Alzheimer's disease and play a key role in signal transduction. Signal transduction pathways are considered to be rich in proteins that might respond to chemical therapy. By integrating information about signal transduction, pyramidal neurons, their genes, and gene products, the query corresponding to our scientific question can provide information relevant to researchers that are looking for drug target candidates that are potentially effective against Alzheimer's Disease.

3 Data Sources

In order to incorporate data from several information sources, it was necessary to convert several exported formats, each into its own RDF bundle. The largest RDF bundle of 200M triples resulted from MeSH associations with PubMed articles. In contrast, there were a number of smaller bundles ranging from 10K to 10M triples. This resulted in a total of approximately 350M triples occupying approximately 20GB when loaded into the RDF repository. In several cases, we extracted only a subset, for example, by selecting only human, rat, and mouse data. Click on [Details] in the table below to view details such as the date of the last extraction.

At the time of publication, the following information sources have been (sometimes partially) incorporated into the knowledge base. This set will continue to be extended in depth (i.e., more complete inclusion of partially represented data sets) and in breadth (i.e., novel data sets):

Allen Brain Atlas (ABA) Allen Brain Atlas is an interactive, genome-wide image database of gene expression in the mouse brain. A combination of RNA in situ hybridization data, detailed Reference Atlases and informatics analysis tools are integrated to provide a searchable digital atlas of gene expression. Together, these resources present a comprehensive online platform for exploration of the brain at the cellular and molecular level. [Details]
Addgene A catalog of plasmids from Addgene [Details]
BAMS The Brain Architecture Management System (BAMS) is designed to be a repository of information about brain structures from different species, and has a set of inference engines for processing the neurobiological data. BAMS contains five interrelated modules: Brain Parts (brain regions, major fiber tracts, and ventricles), Cell Types, Molecules, Relations (between structures from different neuroanatomical atlases), and Connections. [Details]
GALEN GALEN is an advanced terminology of medical concepts for clinical information systems. We imported the GALEN ontology in OWL from CO-ODE. [Details]
NCBI gene_info Information from the gene_info file distributed by NCBI that was imported into OWL. [Details]
Gene Ontology (GO) The Gene Ontology project provides a controlled vocabulary to describe gene and gene product attributes in any organism. GO terms are often used to annotate gene and protein records. [Details]
GOA GO annotations from National Center for Biotechnology Information (NCBI) and European Bioinformatics Institute (EBI). [Details]
HomoloGene Homologene is a system for automated detection of homologs among the annotated genes of several completely sequenced eukaryotic genomes. [Details]
MEDLINE/PubMed PubMed is a service of the U.S. National Library of Medicine that includes over 17 million citations from MEDLINE and other life science journals for biomedical articles back to the 1950s. PubMed includes links to full text articles and other related resources. [Details]
MeSH Medical Subject Headings. 2008 MeSH includes the subject descriptors appearing in MEDLINE/PubMed, the National Library of Medicine (NLM) catalog database, and other NLM databases. [Details]
Neurocommons Text Mining Pilot Protein/gene associations/interactions extracted from Temis software applied to 7% of Medline records. Annotations were captured in RDF using the Neurocommons Annotations Schema. [Details]
Open Biomedical Ontologies (OBO) All Open Biomedical Ontologies (OBO) available from Berkeley Bioinformatics Open-source Projects. [Details]
Science Commons Ontology A bridging ontology, from Science Commons, importing other ontologies used in the prototype, defining classes and relations used to represent gene records and their contents, as well as few items referred to by imported data sources, but not available in a published ontology. [Details]
SenseLab See Experiences with the conversion of SenseLab databases to RDF/OWL. [Details]
SWAN Semantic Web Applications in Neuromedicine [SWAN] is a knowledge base of hypotheses, claims, and evidence in Alzheimer Disease (AD) research, created through a community process to capture the collective scientific insights of the AD field. Not yet public
SKOS Simple Knowledge Organization System (SKOS): specifications and standards to support the use of knowledge organization systems (KOS) such as thesauri, classification schemes, subject heading systems and taxonomies within the framework of the Semantic Web. [Details]

4 Design Decisions

A number of design decisions were made during the construction of the prototype knowledge base. Many of the decisions were pragmatic in nature, as a consequence of the need to implement the solution on a commodity PC within a two-month period for a demonstration at WWW2007.

5 Importing to RDF - Homologene Example

A number of different approaches were used for the conversion of data into RDF/OWL. The most commonly used approach was the use of Lisp code to read text exports of the data and create OWL or RDF documents. We will focus on the example of importing data from Homologene.

The general steps required to import from an existing data source into RDF are:

In the case of Homologene, we start with a text file that contains the exported information. The original tab delimited file is ftp://ftp.ncbi.nih.gov/build54/homologene.data.

Here is a sample of the original file:

99949	9606	727759	LOC727759	113427825	XP_001125931.1
99949	10116	678753	LOC678753	109498373	XP_001053282.1
99949	5833	812783	GeneID:812783	16805082	NP_473111.1
99950	3702	820917	AT3G16650	18401203	NP_566557.1

We are interested in the first 3 fields. The first field identifies the homologous cluster. The second field is the species taxon. The third field is the EntrezGene id. We are only interested in human, mouse, rat, taxon ids: "9606" "10116" "10090".

The Lisp code for the homologene conversion is also available. In the conversion process, we first iterate over the lines in the file, creating a table mapping cluster id to the pairs of taxon id, entrez id in the cluster. This is the variable homologene, created by the function read-homologene. For each of these clusters we will create an individual to represent the cluster e.g for cluster 99949:

  <sciencecommons:orthology_record rdf:about="http://purl.org/science/record/homologene/cluster_r54_99949">
    <sciencecommons:has_homologous_gene_record rdf:resource="http://purl.org/commons/record/ncbi_gene/678753"/>
    <sciencecommons:has_homologous_gene_record rdf:resource="http://purl.org/commons/record/ncbi_gene/727759"/>
    <sciencecommons:has_supporting_evidence rdf:resource="http://purl.org/science/evidence/homologene/cluster_r54_99949"/>

Above is the RDF/XML (see [RDF]) expression of:

@PREFIX homologene: <http://purl.org/science/record/homologene/>
homologene:cluster_r54_99949 sciencecommons:has_homologous_gene_record rdf:resource ncbi_gene:678753 .
homologene:cluster_r54_99949 sciencecommons:has_homologous_gene_record rdf:resource ncbi_gene:727759 .
homologene:cluster_r54_99949 sciencecommons:has_supporting_evidence homologene:cluster_r54_99949 .

Note that we used HTTP URLs to identify Homologene records by prefixing the EntrezGene identifiers (e.g. 727759) with a stem URL, http://purl.org/commons/record/ncbi_gene/. The resulting URL can be usefully resolved with a web browser. The domain purl.org serves Persistent URLs (PURLs), which currently redirect these requests for NCBI gene identifiers to a script at sw.neurocommons.org. If the community wishes to move the service to, for instance, an NCBI page about these genes, they can simply notify the custodians of purl.org. This extra level of indirection protects these identifiers from becoming orphaned as organizations stop existing or change their priorities. These URLs were also used to identify gene information imported from other data sources, automatically linking the Semantic Web representations of these records. For example, PubMesh statements about gene records use these same identifiers for genes, as do the statements from Gene Ontology and SenseLab. This allows for trivial data integration between different resources involving Entrez Gene records.

Also, the individual http://purl.org/science/evidence/homologene/cluster_r54_99949 serves as a link to the "evidence", which is not elaborated in this translation, but would include the BLAST scores and other evidence used to establish the orthology in future work. (see http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=homologene&dopt=AlignmentScores&list_uids=99949)

6 Query

Our scientific question can be summarized as "What genes are involved in signal transduction that are related to pyramidal neurons?". The scientific question can be answered with the following query, which searches for gene names and processes from four data sources within the knowledge base. The data sources include: MeSH (Pyramidal Neurons), PubMed (Journal Articles), Entrez Gene (Genes), Gene Ontology (Signal Transduction). The example query selects the gene name of the genes involved in signal transduction that are related to pyramidal neurons. Some of the complexity in this query comes from the need to capture relevant anatomical and functional detail at the subcellular and molecular level. The portion probing the Gene Ontology queries a set of classes describing processes at the molecular level. Our query employs the SPARQL RDF query language to perform knowledge integration across the sources of the knowledge base. Details on SPARQL can be found in the References.

[Note: The query below will not work verbatim at SPARQL endpoints. We have simplified the actual Banff demonstration query for explanatory purposes in our example below. The Banff demonstration query is discussed in more detail in Named Graphs Section. You can try running the query HERE.].

Please note that the same color (and CSS class) is used to connect the descriptive text in the query with relevant portions of the following figures.

Source (colored) CSS class
PubMesh mesh
Gene Ontology Annotation (GOA) goa
Entrez Gene glbl
Gene Ontology plbl
SELECT ?genename ?processname

  # PubMeSH includes ?gene_records mentioned in ?articles which are identified by pmid in ?pubmed_records .
  ?pubmed_record sc:has-as-minor-mesh mesh:D017966 .
  ?article sc:identified_by_pmid ?pubmed_record .
  ?gene_record sc:describes_gene_or_gene_product_mentioned_by ?article .

  # The Gene Ontology has a set of ?proteins such that foreach ?protein, ?protein ro:has_function [ ro:realized_as ?process ].
  ?protein rdfs:subClassOf ?restriction1 .
  ?restriction1 owl:onProperty ro:has_function .
  ?restriction1 owl:someValuesFrom ?restriction2 .
  ?restriction2 owl:onProperty ro:realized_as .
  ?restriction2 owl:someValuesFrom ?process .
  # Also, foreach ?protein, ?protein has a parent class which is linked by some predicate to ?gene_record.
  ?protein rdfs:subClassOf ?protein_superclass .
  ?protein_superclass owl:equivalentClass ?restriction3 .
  ?restriction3 owl:onProperty dnaGeneProduct:described_by .
  ?restriction3 owl:hasValue ?gene_record .
  # Each ?process (that we are interested in) is a subclass of the signal transduction process.
  ?process obo:part_of go:GO_0007165 .

  ?gene_record rdfs:label ?genename .

  ?process rdfs:label ?processname .

The following shows a few of the results from the query:

gene_record_name processname
Entrez Gene record for human DRD1, 1812 adenylate cyclase activation
Entrez Gene record for human ADRB2, 154 adenylate cyclase activation

The following section describes the RDF data model and how we employed it to make our query possible.

7 Data Model

The data in the knowledge base is modeled in OWL-DL, which has been expressed as RDF triples. Briefly, an RDF triple consists of a subject, predicate, and object. The predicate is also known as the property of the triple. Subjects and objects in the data unify to create an RDF Graph, with subjects and objects as nodes and predicates as edges. For more information about RDF and OWL, see the References section in the Appendix.

Nodes labeled with a leading "_:", e.g. _:activateAdenalCyclase, are called RDF blank nodes [CONCEPTS]. These frequently have machine-generated identifiers and are therefore typically opaque to a human reader (e.g., the set of all nodes that represent protein entities linked to the GO molecular function Adenal Cyclase Activation). Here, for the purposes of explanation, they have been named to convey meaning to the reader. Blank nodes ending in "_1" in this document indicate this blank node is one of many in this class, e.g. _:signalingParticipants_1.

Triples in Solution

Figure 1. Triples in Solution [SVG image PNG image]

Figure 1, Triples in Solution, shows a graphical representation of the triples that compose one solution to the query posed in section 6. Following is a discussion of the origins and intents of those triples:

The application of a commercial text mining tool to neuroscience-related PubMed abstracts results in a set of annotations that link MeSH terms to genes (for more details on MeSH, see the table in Data Sources). An article with PubMed id 10698743 mentions ncbi_gene:1812 and that the corresponding PubMed record has a MeSH term mesh:D017966. The following three triples express this:

subject predicate object
pubmedRec:10698743 sc:has-as-minor-mesh mesh:D017966 .
article:10698743 sc:identified_by_pmid pubmedRec:10698743 .
ncbi_gene:1812 sc:describes_gene_or_gene_product_mentioned_by article:10698743 .

A set of genes or gene products in human bodies are described by ncbi_gene:1812. Here, we call this set _:equiv1812.

_:equiv1812 owl:onProperty dnaGeneProduct:described_by .
_:equiv1812 owl:hasValue ncbi_gene:1812 .

protein:ncbi_gene.1812 has the same extension (members) as the OWL restriction _:equiv1812.

protein:ncbi_gene.1812 owl:equivalentClass _:equiv1812 .

The expression

NamedClass equivalentClass R .
R onProperty SomeProperty .
R hasValue SomeClass

is the RDF representation of an OWL class axiom that says: for all X such that

X SomeProperty SomeClass .

X is a member of the class NamedClass (and vice versa). See OWL Web Ontology Language Semantics and Abstract Syntax Section 4. Mapping to RDF Graphs for a formal treatment of this.

Using our other supplied constant, we note that adenylate cyclase activation, go:GO_0007190, is part of signal transduction, go:GO_0007165. Note: this simplified query matches only processes that are a sub-process of go:GO_0007165; the actual query, described in §9 Named Graphs, looks also for subclasses. The part_of relationships were inferred from the OWL class restrictions described in §7.1 Precomputing Inferences. The class of functions that are realized_as adenylate cyclase activation is here labeled _:activateAdenylCyclase.

go:GO_0007190 obo:part_of go:GO_0007165 .
_:activateAdenylCyclase owl:onProperty ro:realized_as .
_:activateAdenylCyclase owl:someValuesFrom go:GO_0007190 .

There are many possible classes of substance participating in molecular signaling, one of which (called here _:molecularSignalers_1) is defined by the ability to activate adenyl cyclase.

_:signalingParticipants_1 owl:onProperty ro:has_function .
_:signalingParticipants_1 owl:someValuesFrom _:activateAdenylCyclase .

The class of proteins in the intersection of _:signalingParticipants_1 and protein:ncbi_gene.1812 is here abbreviated proteinsubclass:p1812_7190_1, though the actual identifier is proteinsubclass:product_of_ncbi_gene.1812_that_participates_in_GO_0007190_fbc49f20524727a24c7b7effa29bad4a. Note: the Venn diagram reveals that this set is potentially empty, theoretically permitting the query to range over pairs of gene/process that aren't related through any known protein. However, OWL-DL reasoners will not infer new classes, so the proteins in the intersection of ncbi_gene:1812 and the substances participating in molecular signaling is restricted to the set which have already been entered into the knowledge base, e.g. like proteinsubclass:p1812_7190_1

proteinsubclass:p1812_7190_1 rdfs:subClassOf _:signalingParticipants_1 .
proteinsubclass:p1812_7190_1 rdfs:subClassOf protein:ncbi_gene.1812 .

ncbi_gene:1812 and go:GO_0007190 have human-readable labels.

ncbi_gene:1812 rdfs:label "Entrez Gene record for human DRD1, 1812" .
go:GO_0007190 rdfs:label "adenylate cyclase activation" .

The addition of another MeSH record gives us another solution:

pubmedRec:11441182 sc:has-as-minor-mesh mesh:D017966 .
article:11441182 sc:identified_by_pmid pubmedRec:11441182 .
ncbi_gene:1812 sc:describes_gene_or_gene_product_mentioned_by article:11441182 .

7.1 Precomputing Inferences

obo:part_of Rule

Figure 2. obo:part_of Rule [SVG image PNG image]

The demonstration query depends on the existence of an obo:part_of (or rdfs:subClassOf) relationship between any part (i.e. any subclass of any step in the sequence) of molecular signaling, and the general identifier for molecular signaling, go:GO_0007165:

?process obo:part_of go:GO_0007165 .

Triples of this form were generated by a rule, graphically expressed in Figure 2, obo:part_of Rule. The shaded area on the right of the figure shows the OWL restriction which is the antecedent of the rule:

_:subPart owl:onProperty obo:part_of .
_:subPart owl:allValuesFrom _:subClass .
_:subClass owl:onProperty rdfs:subClassOf .
_:subClass owl:hasValue _:parentClass .

The symmetric property for rdfs:subClassOf need not be explicitly modeled because the RDF Schema Specification defines subClassOf, including its transitivity. Note that if _:subClass is a subClassOf _:parentClass, then all members of _:subClassOf are of type _:parentClass (as well as _:subClass):

_:subClass owl:onProperty rdf:type .
_:subClass owl:hasValue _:parentClass .

Because the triple store used does not perform inferencing, these triples have been pre-computed (forward-chained) and inserted into the triple store. This also simplifies the query. If these triples were not pre-computed, the obo:part-of part of the query would be expressed:

?process rdfs:subClassOf ?what .
?what owl:onProperty obo:has_part .
?what owl:someValuesFrom go:GO_0007165 .

and would need to query over a transitive closure of the union of the obo:part-of and rdfs:subClassOf rules.

8 Adding a New Data Source

SenseLab is a collection of relational (Oracle) databases for neuroscientific research that was independently added to the knowledge base after the other data sources. An accompanying document, Experiences with the conversion of SenseLab databases to RDF/OWL, describes the details of adding it to this knowledge base. With this new data incorporated, the example query could be extended to extract data from the new data source, in this case, discovering the names of receptor proteins associated with the genes discovered in the previous query. In an integrative query of this sort, we can use the results as a starting point for more detailed queries of a particular repository, such as in this case SenseLab.

SELECT ?genename ?processname ?receptor_protein_name

  # PubMeSH includes ?gene_records mentioned in ?articles which are identified by pmid in ?pubmed_records .
  ?pubmed_record sc:has-as-minor-mesh mesh:D017966 .
  ?article sc:identified_by_pmid ?pubmed_record .
  ?gene_record sc:describes_gene_or_gene_product_mentioned_by ?article .

  # The Gene Ontology asserts that foreach ?protein, ?protein ro:has_function [ ro:realized_as ?process ].
  ?protein rdfs:subClassOf ?restriction1 .
  ?restriction1 owl:onProperty ro:has_function .
  ?restriction1 owl:someValuesFrom ?restriction2 .
  ?restriction2 owl:onProperty ro:realized_as .
  ?restriction2 owl:someValuesFrom ?process .
  # Also, foreach ?protein, ?protein has a parent class which is linked by some predicate to ?gene_record.
  ?protein rdfs:subClassOf ?protein_superclass .
  ?protein_superclass owl:equivalentClass ?restriction3 .
  ?restriction3 owl:onProperty dnaGeneProduct:described_by .
  ?restriction3 owl:hasValue ?gene_record .
  # Each ?process (that we are interested in) is a subclass of the signal transduction process.
  ?process obo:part_of go:GO_0007165 .

  ?gene_record rdfs:label ?genename .

  ?process rdfs:label ?processname .

  # Foreach ?gene, ?gene senselab:has_nucleotide_sequence_described_by ?gene_record .
  ?gene owl:equivalentClass ?restriction4 .
  ?restriction4 owl:onProperty senselab:has_nucleotide_sequence_described_by .
  ?restriction4 owl:hasValue ?gene_record .

  # Foreach ?receptor_protein, ?receptor_protein senselab:proteinGeneProductOf ?gene .
  ?receptor_protein rdfs:subClassOf ?restriction5 .
  ?restriction5 owl:onProperty senselab:proteinGeneProductOf .
  ?restriction5 owl:someValuesFrom ?gene .

  # Find the labels of all such ?receptor_proteins.
  ?receptor_protein rdfs:label ?receptor_protein_name

yielding another variable in our results:

gene_record_name processname receptor_protein_name
Entrez Gene record for human DRD1, 1812 adenylate cyclase activation D1 receptor
Entrez Gene record for human ADRB2, 154 adenylate cyclase activation NULL

The additional triples this matched in the SenseLab knowledge base connect to the existing data by talking about the same genes, e.g. ncbi_gene:1812.

Additional Triples from SenseLab

Figure 3. Additional Triples from SenseLab [SVG image PNG image]

Figure 3, Additional Triples from SenseLab, shows a subset of the triples provided by SenseLab. Following is a discussion of the origins and intents of those triples:

A nucleotide sequence is also described by ncbi_gene:1812. Here, we call this _:nucleo1812.

subject predicate object
_:nucleo1812 owl:onProperty nucleotideSequence:described_by .
_:nucleo1812 owl:hasValue ncbi_gene:1812 .

The class senselab:DRD1_Gene has the same members as the OWL restriction _:nucleo1812.

senselab:DRD1_Gene owl:equivalentClass _:nucleo1812 .

This _:protGeneProd_1 is defined by being a product of DRD1_Gene.

_:protGeneProd_1 owl:onProperty senselab:proteinGeneProductOf .
_:protGeneProd_1 owl:someValuesFrom senselab:DRD1_Gene .

Our solution is a subclass of _:protGeneProd_1 called senselab:D1.

senselab:D1 rdfs:subClassOf _:protGeneProd_1 .
senselab:D1 rdfs:label "D1" .

9 Named Graphs

In the Banff Demo, the resulting knowledge base partitioned the assertions into groups called Named Graphs. This process basically consists of associating a distinct URI with a connected graph of triples, and then referring to that graph via the URI. At the time of publication, any query would be expected to include SPARQL GRAPH constraints, e.g.:

prefix go: <http://purl.org/obo/owl/GO#>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
prefix owl: <http://www.w3.org/2002/07/owl#>
prefix mesh: <http://purl.org/commons/record/mesh/>
prefix sc: <http://purl.org/science/owl/sciencecommons/>
prefix ro: <http://www.obofoundry.org/ro/ro.owl#>
prefix senselab: <http://purl.org/ycmi/senselab/neuron_ontology.owl#>
prefix obo: <http://purl.org/obo/owl/obo#>

SELECT ?genename ?processname ?receptor_protein_name

  # PubMeSH includes ?gene_records mentioned in ?articles which are identified by pmid in ?pubmed_records .
GRAPH <http://purl.org/commons/hcls/pubmesh> {
  ?pubmed_record sc:has-as-minor-mesh mesh:D017966 .
  ?article sc:identified_by_pmid ?pubmed_record .
  ?gene_record sc:describes_gene_or_gene_product_mentioned_by ?article

  # The Gene Ontology asserts that foreach ?protein, ?protein ro:has_function [ ro:realized_as ?process ].
GRAPH <http://purl.org/commons/hcls/goa> {
  ?protein rdfs:subClassOf ?restriction1 .
  ?restriction1 owl:onProperty ro:has_function .
  ?restriction1 owl:someValuesFrom ?restriction2 .
  ?restriction2 owl:onProperty ro:realized_as .
  ?restriction2 owl:someValuesFrom ?process .
  # Also, foreach ?protein, ?protein has a parent class which is linked by some predicate to ?gene_record.
  ?protein rdfs:subClassOf ?protein_superclass .
  ?protein_superclass owl:equivalentClass ?restriction3 .
  ?restriction3 owl:onProperty sc:is_protein_gene_product_of_dna_described_by .
  ?restriction3 owl:hasValue ?gene_record .
  # Each ?process (that we are interested in) is a subclass or component of the signal transduction process.

  GRAPH <http://purl.org/commons/hcls/20070416/classrelations> {
      { ?process obo:part_of go:GO_0007165 }
      { ?process rdfs:subClassOf go:GO_0007165 }

GRAPH <http://purl.org/commons/hcls/gene> {
  ?gene_record rdfs:label ?genename

GRAPH <http://purl.org/commons/hcls/20070416> {
  ?process rdfs:label ?processname

GRAPH <http://purl.org/ycmi/senselab/neuron_ontology.owl> {
  # Foreach ?gene, ?gene senselab:has_nucleotide_sequence_described_by ?gene_record .
  ?gene owl:equivalentClass ?restriction4 .
  ?restriction4 owl:onProperty senselab:has_nucleotide_sequence_described_by .
  ?restriction4 owl:hasValue ?gene_record .

  # Foreach ?receptor_protein, ?receptor_protein senselab:proteinGeneProductOf ?gene .
  ?receptor_protein rdfs:subClassOf ?restriction5 .
  ?restriction5 owl:onProperty senselab:proteinGeneProductOf .
  ?restriction5 owl:someValuesFrom ?gene .

  # Find the labels of all such ?receptor_proteins.
  ?receptor_protein rdfs:label ?receptor_protein_name

The named graphs help with both provenance and scaling. In the current approach, each RDF bundle is imported into its own named graph. This is useful for a number of reasons. First, we know the source of each named graph, so we can control and review which data sources are being accessed by our queries. Additionally, the association of a named graph with a data source serves as data provenance and can also be employed by schemes that exploit knowledge about the data source to assign confidence measures in a model of trust. For example, one of the knowledge base data sources resulted from text mining experiments to find protein associations. Users of the knowledge base can choose to view this evidence of association differently than the associations provided from a protein-protein interaction database. Also, named graphs support scaling by making it possible to update selected parts of the knowledge base, for example when the data source has new information or related ontologies are changed.

10 Opportunities for further development

The knowledge base was initially designed for the purposes of a live demo. It also provided a basis for early work on the Neurocommons, where its development continues. Some design choices were made to favor simplicity and maximal performance, including the use of a central triple store, and the design of the data and queries. Many of the choices were guided by the desire for transparency for a broader audience of biomedical informaticists. Several areas of possible improvement are noted here:

There are also a number of open issues that should be addressed in future work:


A RDF Sources

A table of the RDF sources used to create the Knowledge base:

RDF bundle name Last modifiedSize Description RDF conversion by Terms
aba-2007-08-07.tgz 22-Sep-2007   51M SC's extract of Allen Brain Atlas metadata from their Web site. Web site was read on 26 Feb 2007 or shortly before SC terms of use
addgene.ttl 16-May-2007 1.1M Addgene catalog (tab-delimited file) SC provided to Science Commons by Addgene
bams-from-swanson-98-4-23-07.owl 23-Apr-2007 5.6M BAMS HCLSIG/NIST released without contract
galen.tgz 22-Sep-2007 1.9M Galen from co-ode.org - released without contract
gene-owl.tgz 08-May-2007 7.7M Select fields from Entrez Gene records HCLSIG/SC NCBI Copyright and Disclaimers
gene-pubmed.ttl.tgz 08-May-2007 1.5M Entrez Gene Extract from ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_info.gz HCLSIG/HP/SC NCBI Copyright and Disclaimers
goa-in-owl.tgz 16-May-2007 73M GO annotations from National Center for Biotechnology Information (NCBI) and European Bioinformatics Institute (EBI) HCLSIG/SC NCBI Copyright and Disclaimers; EBI terms of use
homologene.tgz 16-May-2007 626K Homologene HCLSIG/SC NCBI Copyright and Disclaimers
(contact Medline for use terms)
16-May-2007 758M List of all associations of MeSH headings to papers indexed by Medline extracted from 2007 Medline baseline distribution HCLSIG/SC License Agreement to Lease NLM Databases in Machine-Readable Form - see below
(contact Medline for use terms)
16-May-2007 670M Extracted from 2007 Medline baseline distribution HCLSIG/SC see below
mesh-qualified-headings.ttl.gz 30-Apr-2007 13M NLM 2007 MeSH descriptor/qualifier pairs HCLSIG/SC MeSH MOU
mesh-skos.tgz 16-May-2007 13M NLM 2007 MeSH van Assem et al/SC MeSH MOU
mesh07-eswc06.rdfs 28-Jun-2007 2.2K van Assem et al's ontology (used by output of MeSH to SKOS conversion) van Assem et al released without contract
neurocommons-text-mining.tgz 05-May-2007 24M Neurocommons text mining pilot - extracted from Temis software applied to 7% of Medline records SC released without contract
obo-all.tgz 22-Sep-2007 36M All OBO ontologies BBOP released without contract
obo-in-owl.tgz 16-May-2007 2.6M selected OBO ontologies, downloaded ~21 April 2007, augmented with inferred relations HCLSIG/SC released without contract
sciencecommons.owl 28-Jun-2007 19K A bridging ontology, from Science Commons, importing other ontologies used in the prototype, defining classes and relations used to represent gene records and their contents, as well as few items referred to by imported data sources, but not available in a published ontology. HCLSIG/SC released without contract
senselab.tgz 16-May-2007 216K From Yale Senselab HCLSIG/Yale released without contract


Science Commons (SC), Berkeley Bioinformatics Open-source Projects(BBOP), Health Care and Life Sciences Interest Group (HCLSIG), National Institute of Standards and Technology (NIST), Hewlett Packard (HP)

B References

OWL Web Ontology Language Overview,
Deborah L. McGuinness and Frank van Harmelen, Editors,
W3C Recommendation, 10 February 2004,
http://www.w3.org/TR/2004/REC-owl-features-20040210/ .
Latest version available at http://www.w3.org/TR/owl-features/ .
Resource Description Framework (RDF) Model and Syntax Specification,
Ora Lassila, Ralph R. Swick, Editors,
World Wide Web Consortium Recommendation, 1999,
Latest version available at http://www.w3.org/TR/REC-rdf-syntax/.
Resource Description Framework (RDF): Concepts and Abstract Syntax,
G. Klyne, J. J. Carroll, Editors,
W3C Recommendation, 10 February 2004,
http://www.w3.org/TR/2004/REC-rdf-concepts-20040210/ .
Latest version available at http://www.w3.org/TR/rdf-concepts/ .
SPARQL Query Language for RDF, A. Seaborne, E. Prud'hommeaux, Editors, W3C Recommendation, 15 January 2008, http://www.w3.org/TR/2008/REC-rdf-sparql-query-20080115/ . Latest version available at http://www.w3.org/TR/rdf-sparql-query/ .
Turtle - Terse RDF Triple Language,
W3C Team Submission, 14 January 2008,
http://www.w3.org/TeamSubmission/turtle/ .
Alzforum and SWAN: The Present and Future of Scientific Web Communities,
Clark T and Kinoshita J.,
Briefings in Bioinformatics 2007;8:163-171 doi:10.1093/bib/bbm012.

C Additional Resources

The knowledge base has been installed at several locations. At the time of this writing, these locations provide SPARQL query access, however, it is not guaranteed that the endpoints at these address will persist, or continue to serve the knowledge base described in this note:

Below are a few visual interfaces that make it possible to browse the results of a search on the knowledge base:

Entrez Neuron was developed by the SenseLab team as a graphical user interface for querying the SenseLab ontologies.

We used the open source edition of the Openlink Virtuoso repository from http://sourceforge.net/projects/virtuoso/.

The actions and scripts that were used to create the knowledge base on a commodity PC have been documented by several HCLSIG members. The necessary instructions and scripts that were used will be listed here as completely as possible:

The following resources may be of interest for future work:

D Acknowledgements

In memory of our friend and colleague William Bug, Ontological Engineer.

Special thanks to: Alan Ruttenberg (Science Commons) who coordinated the assembly, conversion, and deployment of the data sets and ontologies and Susie Stephens (Eli Lilly) who coordinated the BioRDF task force. Together they presented the initial version of the knowledge base at a WWW2007 Banff workshop.


Many contributed to the development, documentation and validation of the knowledge base, as well as the thinking behind it.

Mikail Bota (USC) who kindly provided the BAMS database for our use and John Barkley (NIST) converted it to RDF. Huajun Chen (Zhejiang University), Matthias Samwald (Yale Center for Medical Informatics; DERI Galway; Semantic Web Company), Alan Ruttenberg, and Kei-Hoi Cheung (Yale Center for Medical Informatics) participated in the the SenseLab RDF Conversion. Members of the SWAN team: Tim Clark, Paolo Ciccarese, June Kinoshita, Gwen Wong, and Elizabeth Wu contributed the SWAN data source. June Kinoshita, Gwen Wong, Elizabeth Wu, Don Doherty (Brainstage Research Inc.), William Bug (School of Medicine, UCSD), and Alan Ruttenberg worked on the neurogenerative disease use cases. Ray Hookaway (HP) provided digests from Entrez Gene that were more easily converted to RDF. Jonathan Rees (Science Commons) did the RDF conversions of Addgene, Pubmed to Gene, Medline, and MeSH, the Neurocommons text mining pilot, and compiled the data source and licensing information for this document. Alan Ruttenberg did the RDF conversion of Entrez Gene records, GO Annotations, Allen Brain Atlas, Homologene, wrote the Science Commons ontology. Alan Ruttenberg and Matthias Samwald wrote the SPARQL queries described in this document. Chris Mungall (NCBO) wrote the converter that produced the OWL versions of the OBO ontologies and consulted on matters of ontology.

Eric Neumann (Clinical Semantics Group) produced the Exhibit visualization. Alan Ruttenberg developed the Google Mouse prototype, with contributions from Mike Travers (CollabRX), Brian Gilman (SciLink), and Tom Stambaugh (Zeetix). Don Doherty, Matthias Samwald, Holger Stenzorn (DERI), M. Scott Marshall, and Eric Prud'hommeaux have presented this work at conferences.

Barry Smith (State University of New York at Buffalo, USA) provided advice on ontology work and led the development of the Basic Formal Ontology, which inspired all ontology work related to the knowledge base.

William Bug (School of Medicine, UCSD), Michel Dumontier (Carleton University), and Holger Stenzorn (DERI) reviewed and gave detailed comments on an initial draft of this note. Alan Ruttenberg Jonathan Rees and Susie Stephens, reviewed and contributed to several versions of the document. Susie Stephens coordinated the BioRDF task force, worked on presentations of the work, and wrote the introduction to this document. M. Scott Marshall (University of Amsterdam) and Eric Prud’hommeaux (W3C) edited and coordinated the production of this note. Eric Prud’hommeaux created the figures.

We would like to offer special thanks for organizations which gave contributions of equipment and service. Through Ray Hookway and Jeannine Crockford, Hewlett Packard donated two machines for a period of six weeks during the demo. Science Commons hosted the the prototype during development and continues to host and develop a knowledge base derived from the prototype as part of the Neurocommons. MIT CSAIL hosts Science Commons and provided computer and networking infrastructure.

Kingsley Idehen, Orri Erling, Ivan Mikhailov, Mitko Iliev, Patrick van Kleef and Anton Avramov from Openlink Software provided rapid technical support including several custom builds of the Virtuoso triple store to address early performance issues, making it possible to develop the prototype on an aggressive schedule. Evren Sirin from Clark and Parsia provided support for the Pellet OWL reasoner.

In addition to data sources that were incorporated into the prototype, other data that did not make it in was provided by Judith Blake (MGD) an Simon Twigger (RGD), and Colin Knep (Alzforum)