This work focuses on automated incremental development of biological networks. The incremental approach is demonstrated on two use cases. First, a simple plant defence network created manually is extended in two incremental steps, yielding the final model with 183 relations. Second, a complex published network of plant defence response is incrementally updated with 104 new relations automatically extracted from recently published articles. The results show that using the demonstrated incremental approach it is possible to automatically recognise new knowledge about the selected biological relations published in recent scientific literature.
COBISS.SI-ID: 28591655
A conventional way of manual construction of a dynamic model can be accompanied by additional steps, which can speed up and enhance model construction. In our work, we have developed a new methodology for constructing biological models based on domain knowledge and literature. We have applied this methodology to the construction of the defence response model in plants. Using solely manual knowledge engineering was not feasible due to the complexity of a plant defence system. For this reason, the laborious manual model creation was complemented by additional automatised and semi-automatised steps. These steps address the model structure, which was enhanced by means of natural language processing techniques, as well as model dynamics, where parameter optimisation was guided by constraints defined by the domain experts.
COBISS.SI-ID: 28827175
The paper describes a prototype semantic data mining system called g-SEGS, which uses ontologies as background knowledge in the learning process. The system is a generalization of an existing system SEGS, which was successfully used in the field of functional genomics, but cannot be applied to other fields. g-SEGS is implemented as a web service and integrated into Orange4WS, a visual programming environment for data mining. In addition, the paper describes how to formulate the problem of semantic data mining in the inductive logic programming system Aleph. Both approaches are experimentally evaluated on two real-life biological domains.
COBISS.SI-ID: 28293671