This paper is a summary of a recent work accepted (as a long paper) for publication in the tenth international conference on Data Integration in Life Sciences (DILS 2014) [1]. We briefly describe in the last section ongoing work for improving the rule-based prediction process. Increasing amounts of biomedical data provided as Linked Open Data (LOD) offer novel opportunities for knowledge discovery. LOD are represented in a standard format, partially integrated, and offer connections with domain knowledge available in semantic web ontologies. Relational data mining methods such as ILP are good candidates to consider together LOD and domain knowledge awareness. We propose in this paper an approach for collecting and mining LOD, using ILP, with the goal of characterizing and predicting genes responsible for a disease. An integration step enables to select and link together relevant pieces of LOD. It results from this integration a graph that is subsequently mined us-ing ILP. For this real-world use case, we design ILP experiments on two subsets of relational descriptors of genes responsible for intellectual dis-ability. We evaluate ILP results and assess the contribution of domain knowledge. Our ongoing efforts explore how the combination of rules coming from distinct theories can improve the prediction accuracy.