Because current databases of human metabolic network are far from complete. Which would lead to a false-negative/positive prediction. On the contrary, our method based on a global distance measure appeared to be more tolerant of incomplete data. Even when we deleted 20% edges of metabolic network, the AUC value had only a slight decline. Our strategy was proved successful in prioritizing known metabolite for 71 diseases with an AUC value up to 0.895. Especially, it had good performance on metabolic-related diseases. Secondly, might be more important, our PROFANCY method sufficiently exploited the functionally modular information of metabolic network. The metabolic network was divided into different metabolic pathways and the metabolites in the same pathway were strongly functionally related. To fully exploit the functional modularity information of metabolic network, we added functional pathway nodes to the metabolic network. The functional pathway nodes would improve the performance by enhancing the connectivity between metabolites related to the same disease, especially for the disease whose metabolites belonged to different pathways. As we mentioned above, two functional pathway nodes enhanced the connectivity of kynurenate and pipecolic acid which were both related to malaria but belonging to different pathways. The results showed that this strategy had effectively Masitinib improved the performance–three metabolites of malaria were all ranked in top 10% and the AUC for immunological diseases increased from 0.832 to 0.961. The functional pathway nodes also contributed to the robustness of PROFANCY. They could maintain a part of functional relationships between disease metabolites in the incomplete metabolic network. The AUC could achieved to 0.8 even when we removed 70% edges of metabolic network, but this value would declined to 0.65 without functional pathway nodes. We also noticed that there were some limitations of our PROFANCY method. At first, our method depended on the topology of the metabolic network, so the low-quality and incompleteness of reaction information of KEGG or EHMN database might limit its performance. Especially, there were no organ-specific reaction and pathway structure data available currently. Although the PROFANCY could perform well in the incomplete network, the performance could be further improved after more complete and specific reconstructions of metabolic network. Secondly, our result is limited to diseases with known metabolites from the HMDB database and the number of known metabolites might have influence on the performance. Integrating multiple metabolite data sources and availability of well-annotated metabolic pathway may overcome this limitation. A combination of these effects likely underlies the ability of this protein family to influence cell fate and malignant transformation. Depending on context, these proteins may behave as either tumor suppressor genes or as facilitators of oncogenesis as in invasive breast cancer and synovial cell sarcoma, respectively. In Drosophila, Groucho is a master transcriptional regulator expressed ubiquitously throughout development and influences multiple key developmental processes.