Ontologies

Always UnderConstruction

These are some of the ontologies that i have either built or in which I've had a hand in building. I hope I have given credit where appropriate. Many of these ontologies may be found in the Manchester OWL repository.

• the original TAMBIS ontology (TaO) captured biology and bioinformatics. A description of a class of bioinformatics instances is in effect a query against the resources that hold those instances. TAMBIS was a tool that used this approach to integrate a bunch of bioinformatics resources. the original TaO was represented in GRAIL, long before OIL, DAMIL+OIL or owl existed. the bulk of this ontology was built by Pat baker and it is the first ontology upon which I worked. The GRAIL encodings of the TAMBIS ontology are available:
  
  1. The big model that covers Proteins, nucleic acids, their components, function, process, location, publishing and much more.
  2. the baby model that covers only the protein subset of the big model. This was used for the "fully functioning" version of the TAMBIS system.
  3. the reconciled model that is the merged version of the big and baby TAMBIS ontologies.
  The later versions of TAMBIS used a translation of the original Baby TaO in GRAIL into the later DAML+OIL language. There is also an OWL (Web Ontology Language) version of this ontology.
  
  1. The DAML+OIL version of Baby TaO.
  2. The OWL version of Baby TaO.
• TAMBIS TaO re-worked in OIL, etc. the TaO was re-worked in OIL, by hand and without the use of a reasoner to check it out. Several translations into other representations also introduced errors. It has some very peculiar stuff in it now. It has been used as a test for "explanation" in DL reasoners.
• The Pizza ontology forms the core of the Manchester Pizza Tutorial. the ontology is an introduction to the Web Ontology Language (OWL). the ontological content of the ontology is light, but the tutorial is designed to touch almost every aspect of the language and very much concentrates on the role of automated inference and OWL. the idea originated with Angus Roberts, Chris Wroe and myself. It was, however, who did all the real work.
• The pizza example is good for teaching the OWL language, but doesn't teach much ontological modelling. Given the restraint that we want a familiar, accessible example that highlights ontological issues means the choice of topic is difficult. In preparation for the Semantic Web authoring tool (SWAT) project, my former colleague Andrew Gibson suggested the topic area of travel. This obviously implies land, oceans, countries, transport, journey, route, holiday, etc. etc. it also needs lots of individuals. Interestingly, it will also need some account of time. So, it makes a good example. The latest Travel Ontology is available. I'm trying to keep versions of the ontology. At the moment it is a horrible mess..
• the amino Acids ontology is an example of the technique of normalisation. the twenty amino acids are simply described as "amino acids" and all the characteristics of the molecules, such as polarity, hydrophobicity, etc. pulled out as some form of "quality" and represented as a value partition (sometimes with a little sleight of hand). All the categories of amino acid used by biochemists can then be represented as defined classes and the automated reasoner infer the rather tangled heirarchy. Such classes would include Polar amino Acid; Positive amino acid; large positive amino acid; small polar amino acid; small amino acid; etc. etc. the original of this ontology was built with Phil Lord.
• The myGrid service ontology is used to describe the inputs, outputs and tasks performed by the various services used by tools such as Taverna. this ontology was originally built by Chris Wroe and now maintained by Katy Wolstencroft and Franck Tano.
• an ontology of the protein phosphatases was built by Katy wolstencroft as part of her Ph.D. it describes the protein phosphtases in terms of their domain structure. a protein sequence analysed by a tool such as InterProScan can be converted into an OWL instance with facts asserting which domains its sequence contains. Such instances can then be classified against the ontology; new putative protein phosphatases have been found by this method. This work has been taken up by Rachel Brenchley as part of her Ph.D. and she has used the ontology to analyse and compare the protein phosphatases implied by the genomes of three parasites.
• an ontology representing the Periodic table of the elements. this was an exercise, initially to explore the use of datatype properties. the challenge was to represent the kind of physicochemical properties known by Dimitry Mendelev in the 1860's and then write definitions that would infer the "groups" seen within the periodic table. it almost works. the ontology has one major distinction: We need to talk about the particles themselves, atoms and ions, but we also have to talk about amounts of those particles -- moles of atoms. After all, an atom doesn't conduct electricity, but a mole of atoms does so. The atoms, ions, moles of atoms and ions are used to make a lot of salts. From these various groups are inferred (almost). simon Jupp also put a lot of hard work in to this ontology.
• Atoms ontology is a follow up to the Periodic Table Ontology that classifies atoms just by the electrons appearing in their valence electron shell.
• An ontology of sub-atomic particles is a simplistic representation of atomic structure. Its creation was stimulated by the work on the periodic table (above). ?Atoms are defined by the numbers of their protons. Scope of an ontology is always a fraut issue, yet chemists are, by and large, not really interested in anything other than electrons; protons determine the type of atoms; neutrons have some interesting effects; any finer granualrity than electron, proton and neutron is very much out of scope. So, this ontology was simply done to demonstrate something not to do for chemistry. the ontology is based on physics up to the age of sixteen and Wikipedia; I make no apologies.
• this is an ontology of a family history. it was created in a response to a question of how much inference OWL2 can make of the classic family relationships. this ontology uses a complex property hierarchy, property characteristics and sub-property chains to make a lot of inferences over some 450 individuals from my own family. I've only asserted parent relationships and some sibling relationships. there are some weaknesses that stop this being the good example it could be -- the inability to make a property irreflexive when it is used in a sub-property chain makes myself be my own cousin.... there are a couple of other restrictions, but the ontology is still powerful. the addition of a few rules would tidy it up completely and this will be done real soon now. Some defined classes can also draw the correct inferences.
• This is a cube ontology; It is a cube of five by five by five individual cubes, each of the class "cube". facts are asserted as to which other cube is either above, to the left or front of any given cube. A property heirarchy and a series of sub-proeprty chains is used to infer facts about which other faces are adjacent to a given cube; the edges that touch and the corners that touch. all this is done with the assertion of a maximum of three facts per cube. unfortunately, the different "routes" one can make from one face to another adjacent face stop it all working.
In an M.Sc. project, Eleni Mikroyannidi (with Alan Rector) used a selector pattern to abstract over a portion of the Foundational Model of Anatomy and greatly reduce its size whilst retaining all the information. We presented this work at Bio-Ontologies 2009 in Stockholm. The supplementary material for the anatomy abstraction is also available.
• I've made an anatomy of flowers. the task here is to accommodate the variation seen in flowers from grassses to roses. I haven't got as far as grasses yet, but it is an interesting exercise to build a model with a strict semantics to a domain with such massive variation. the ontology lacks comments and natural language definitions. it is what it is.
• A Comparative genomics ontology (CGO was created by Andrew gibson as part of the ComparaGRID project. It describes physical entities such as chromosomes and models of those entities such as maps and sequences. through this mechanism, a variety of differing maps of the same entity can be accommodated. A version of the CGO to which the ARKdb Farm Animal database has been mapped as part of the ComparaGRID project is available as a CGO application ontology.
• The Cell Cycle Ontology has been a collaborative effort led by the Plant systems biology group at VIB in which Mikel Egaña Aranguren has participated during his Ph.D. The CCO is a central component of the BioGateWay, an RDF based semantic systems biology resource.
• Eleni Mikroyannidi and Alan Rector did some work on abstracting the FMA and we wrote the work up for a paper at the bio-Ontologies meeting at ISMB 20o9.
• as part of the OntoGenesis network, we have done an experiment in collaborative ontology building. we did a normalisation of the Cell Ontology from OBO. The normalised ontology has a cell ontology that references gO, PATO species taxonomy and anatomy. the various normalised CTO resources can be found here.