search

sivarajankumar / phenotype-ontologies

Automatically exported from code.google.com/p/phenotype-ontologies
0 stars 1 forks source link

readme

Phenotype Ontologies

This repository aggregates together phenotype ontologies and logical definitions from a variety of different species. See:

Mungall, C. J., Gkoutos, G. V., Smith, C. L., Haendel, M. A., Lewis, S. E., and Ashburner, M. (2010). 
Integrating phenotype ontologies across multiple species. 
Genome Biology 11, R2. 
Available at: http://dx.doi.org/10.1186/gb-2010-11-1-r2

The directory structure follows the standard one recommended for ontology projects:

src/ tools/ ontology/ docs/

Most of the useful material is in the ontology directory.

Use of svn:externals

We use svn:externals to make working with multiple ontologies easier. Whilst every ontology can be obtained from the web via it's standard ontology URI (e.g. http://purl.obolibrary.org/obo/pato.owl), it is easier to have local file access in a standard place.

The current structure is:

src/ ontology/ external/ anatomy/ cell/ hpo/ ...

Note that none of the directories below "external" are actually in this repository - they are dynamically linked.

See the README.txt in the externals directory for more details.

Tools directory

The tools directory contains ad-hoc scripts mostly for performing translations on the logical definitions files

Ontologies

The directory structure of the ontology directory attempts to mirror the structure of the PURLs. So we have:

mp.owl mp.obo mp/ mp-equivalence-axioms.owl mp-equivalence-axioms.obo hp/ hp-equivalence-axioms.owl hp-equivalence-axioms.obo ...

Note that the primary location of the mp equivalence axioms is here - however, the hpo svn repo (linked via svn externals) is primary for the hp logical definitions.

Note that the src/ontology directory contains a catalog-v001.xml. This is used by Protege 4 and by OWLTools to map ontology global URIs to locations within the local filesystem.

Makefile

Aggregating information is controlled entirely by the Makefile in the src/ontology/ directory.

To rebuild a target, just type

make

You may need owltools installed.

Entity matching

The hp-mp/ directory contains the results of the automated lexical entity matching between HPO and MPO. We also manually maintain a file of disjoints that override the results of the entity matching, where we know the label match does not translate to a real equivalence.

We also provide an obo and owl file of these matches. We treat these as equivalence axioms. This is potentially problematic, due to differences between mouse and human.

Importer ontologies

We provide a number of importer ontologies. These contain no axioms of their own, only owl:import declarations to other ontologies.

For more on the concept of importers, see:

http://wiki.geneontology.org/index.php/Ontology_extensions

There are two main importers:

Both import human and mouse phenotype ontologies, together with equivalence axioms and connected ontologies.

The difference is that the first equates phenotypes with qualities inheres in entities, the second equates phenotypes with bundles of qualities inhering in entities

Different representations of phenotypes

The original template of

?Phenotype EquivalentTo ?Quality and inheres_in some ?Entity

does not work well for HPO because it has many phenotypes that are bundles of qualities.

We therefore explore an alternative representation:

?Phenotype EquivalentTo has_part some (?Quality and inheres_in some ?Entity)

i.e. the phenotype is a collection of qualities.

This allows composite phenotypes. E.g.

?Phenotype EquivalentTo has_part some (?Quality1 and inheres_in some ?Entity1)
         and has_part some (?Quality2 and inheres_in some ?Entity2)

This is called the "subquality" model. Files are named "-equivalence-axioms-subq.owl"

Use of Uberon

Uberon classes subsume species-specific classes. Specifically, Uberon renders the species AO classes as taxonomic equivalents. For example:

   'mammary gland (FMA)' EquivalentTo: 'mammary gland (generic)' and part_of some NCBITaxon_9606

See http://uberon.org and http://genomebiology.com/2012/13/1/R5/abstract

For the purposes of phenotype analysis, we can use a stronger interpretation:

   'mammary gland (FMA)' EquivalentTo: 'mammary gland (generic)'

Due to the fact that we mix Uberon and ssAO classes in the logical definitions, this is necessary, although it may lead to some incorrect inferences (such as the location of mammary glands in mice)