Candidate Gene Prioritization Using Network Based Probabilistic Models

We present a new gene prioritization method that learns a probabilistic knowledge model from a knowledge base and free text research documents and exploits it to prioritize candidate genes. The knowledge model is represented by a network of associations among the domain entities (e.g., genes) and is extracted from a domain knowledge base (e.g., protein-protein interaction database) or a corpus of text documents (e.g., PubMed documents). This knowledge model is then used to perform probabilistic inferences and applied to the task of candidate gene prioritization. We evaluate our new method on five diseases and show that it outperforms a recently described network based method for candidate gene prioritization.