Development and application of ontologies for biological applications
von Jürgen Dönitz
Datum der mündl. Prüfung:2016-01-27
Erschienen:2016-02-08
Betreuer:Prof. Dr. Gregor Bucher
Gutachter:Prof. Dr. Gregor Bucher
Gutachter:Prof. Dr. Edgar Wingender
Gutachter:Prof. Dr. Stephan Waack
Gutachter:Prof. Dr. Burkhard Morgenstern
Gutachter:Prof. Dr. Ulrich Sax
Gutachter:Prof. Dr. Ernst A. Wimmer
Dateien
Name:thesis_doenitz.pdf
Size:7.90Mb
Format:PDF
Zusammenfassung
Englisch
In the century of information technology automatic processing of knowledge is evolving to be a crucial challenge. However, it is usually not obvious to extract information from the syntax of texts. Hence, ontologies are an important pillar by defining entities of a knowledge domain and setting them in relation to each other. The multitude of facts within a field is represented as a collection of single statements following a strict syntax. This structured representation of knowledge allows for automatic processing of the information. Within the iBeetle project, more than 5,000 genes were knocked down by RNAi in Tribolium castaneum in a large-scale screen, and the resulting phenotypes were annotated and compiled in the database of the iBeetle-Base website. I developed the Tribolium Ontology (TrOn), a compilation of the morphological structures found in Tribolium castaneum, in order to enable consistent annotation and machine-readable description of the observed phenotypes as well as a user-friendly search function. TrOn supports to search for defects in general anatomical structures like “leg”, although the annotation may have pointed to a substructure, e.g. the coxa, a part of the leg. With around 1,000 terms the ontology covers the morphological structures visible from the outside and some selected internal ones. All structures are modeled for the main developmental stages larva, pupa and adult, regardless of their occurrence in the iBeetle phenotype screen. There are three sets of entities within TrOn defining its architecture: concrete, abstract and mixed classes. Concrete ontology classes represent dissectible anatomical structures of a specific developmental stage. Abstract ontology classes stand for umbrella terms, e.g. “wing”, that rather stresses the concept of a biological function and are independent of the developmental stage. Anatomical structures, which are only present in a single developmental stage, and thus represent a dissectible structure as well as a biological concept, are listed in the set of mixed classes. Initially, ontologies are data stored in a file. Tools are required to process these data, to edit and to visualize them, and also to enable their use as input for statistical calculations. I developed the service Ontology Based Answers (OBA) to offer ontologies and their semantic information to other applications. The OBA service provides access to the ontologies over a network interface. The client of the OBA service uses this 3 Abstract interface to load the connected entities in the background only when required while traversing the ontology. The service is extended by plugins implementing ontology- specific knowledge such as annotation guidelines of the ontology. The OBA service enables an application developer, who is unfamiliar with the semantic of ontologies, to quickly enhance applications similar to iBeetle-Base with the information provided by the ontologies. The ontology TrOn, the ontology service OBA, and the search function of iBeetle-Base, which I developed in this work, demonstrate the benefit of ontologies for biological applications.
Keywords: bioinformatics; ontology; Tribolium castaneum; OBA ontology based answers; iBeetle-Base