Rapid progress in high-throughput biotechnologies (e. g. microarrays) and exponential accumulation of gene functional knowledge makes it promising for systematic understanding of complex human diseases at the functional modules level. Current modular categorizations can be defined and selected more specifically and precisely in terms of both biological processes and cellular locations, aiming at uncovering the modular molecular networks highly relevant to cancers. Based on Gene Ontology, we identifed the functional modules enriched with differentially expressed genes and characterized by biological processes and specific cellular locations. Then, according to the ranking of the disease discriminating abilities of the pre-selected functional modules, we further defined and filtered signature modules which have higher relevance to the cancer under study. Applications of the proposed method to the analysis of a prostate cancer dataset revealed insightful biological modules.