Bridging the explanatory gap between pathology and molecular interactions : a semantic approach for the representation of knowledge about cancer

I explore the way that the representation of knowledge about cancer meets two aspects: explanation and heterogeneous data integration. I will question the possibility of a successful explanation of pathological phenomena on the level of the molecular interactions. By way of an example, I will discuss the case of breast cancer. Finally, I will argue for an integrative approach to pathology, which is not reducible to the molecular level only. For, an explanation of complex multilevel process in terms of molecular interactions alone turns a loss in its explanatory power. Hence, it is deficient in two ways: 1) it is not comprehensible for the human understanding of cancer 2) it hampers a representation of knowledge about cancer in the Life Sciences Knowledge Databases. In my argument I oppose the position according to which problem in the representation of complex biological processes is due to our inability to identify the relevant functional units. I claim that neither a complete knowledge about the components or some functional units involved in the process, nor mere mathematical models of systems biology can offer a fully comprehensible solution for the linkage of the different levels in pathology. The molecular classification of cancer deals with data such as whole genome expression profiles, searching for related irregularities, overexpression, downregulation or mutation of genes and its effects on their products. Seemingly, it is the best candidate for an accurate scientific explanation of cancer. However, if we stay just on the level of gene expression and molecular interactions, without connecting them with other levels of observations that are characterized as relevant for an explanation, we could lose a criterion of significance among billions of genomic data. In order to make sense of data, there is an essential need for an organised representation of existing knowledge. Moreover, there is a need to connect and keep track of different levels of information and reasoning from experimental and clinical data to biomedical claims about disease. Integration of molecular oncology with pathological data is necessary. Only by these means can we have not only a comprehensible representation and explanation of different levels involved in cancer, but also a useful tool that can help us to direct our interest and distinguish what is a relevant unit of explanation. A convenient solution can be found in a semantic approach which uses formal ontologies (*), connecting them into a relational logical structure. It opens a field for a comprehensible representation of pathology. Moreover, the semantic approach can deal with the diverse pragmatic and research interests. In this way, a disordered picture in the field of cancer research could be settled. For, an advantage that the semantics brings in dealing with the complex phenomena, such as cancer, consists in its flexibility and capability to connect different levels of explanation. *)I speak about ontology in an epistemological manner, which has to do with the issue of the classification and representation of knowledge in the biological databases