Affiliations: [a] Department of Library and Information Science, Yonsei University, 50 Yonsei-ro, Seodaemun-gu, Seoul, Republic of Korea | [b] Institute of Convergence, Yonsei University, 50 Yonsei-ro, Seodaemun-gu, Seoul, Republic of Korea
| [c] Department of Neurology, Massachusetts General Hospital Stroke Research Center, Harvard Medical School, Boston, MA, USA
Correspondence:
[*]
Correspondence to: Min Song, Department of Library and Information Science, Yonsei University, 50 Yonsei-ro, Seodaemun-gu, Seoul, Republic of Korea. Tel.: +82 2 2123 2416; Fax: +82 393 8348; E-mail: [email protected].
Abstract: Late onset Alzheimer’s disease (AD) and Parkinson’s disease (PD) are mostly “sporadic” age-related neurodegenerative disorders, but with a clear genetic component. However, their genetic architecture is complex and heterogeneous, largely remaining a conundrum, with only a handful of well-established genetic risk factors consistently associated with these diseases. It is possible that numerous, yet undiscovered, AD and PD related genes might exist. We focused on the ‘gene’ as a mediator to find new potential genes that might have a relationship with both disorders using bio-literature mining techniques. Based on Entrez Gene, we extracted the genes and directional gene-gene relation in the entire MEDLINE records and then constructed a directional gene-gene network. We identified common genes associated with two different but related diseases by performing shortest path analysis on the network. With our approach, we were able to identify and map already known genes that have a direct relationship with PD and AD. In addition, we identified 7 genes previously unknown to be a bridge between these two disorders. We confirmed 4 genes, ROS1, FMN1, ATP8A2, and SNORD12C, by biomedical literature and further checked 3 genes, ERVK-10, PRS, and C7orf49, that might have a high possibility to be related with both diseases. Additional experiments were performed to demonstrate the effectiveness of our proposed method. Comparing to the co-occurrence approach, our approach detected 25% more candidate genes and verified 10% more genes that have the relationship between both diseases than the co-occurrence approach did.
