Evaluating relatedness of gene sets

This paper proposes an approach to evaluate the relatedness between two genes and within a set of genes belonging to coherent organizations. Specifically, we consider families and pathways. Our method is grounded in the ontological annotation of a specialized corpus of texts consisting of gene summaries that condense the description of functions and processes in which genes are involved. First, we derive a set of descriptor terms by annotating each summary with different ontologies. Then, the relatedness between two genes is measured by the overlapping of their descriptor terms. To explore the relatedness of genes within a coherent organization, we evaluate the relatedness for all the possible pair-wise combination of genes belonging to that organization. This results in a discrete variable distribution whose statistical parameters give information about how genes relate coherently within organizations.