Extraction of Gene-Environment Interaction from the Biomedical Literature

Proceedings of the The 8th International Joint Conference on Natural Language Processing, pages 865–874, Taipei, Taiwan, November 27 – December 1, 2017 c⃝2017 AFNLP Extraction of Gene-Environment Interaction from the Biomedical Literature Jinseon You Jin-Woo Chung Wonsuk Yang Jong C. Park∗ School of Computing Korea Advanced Institute of Science and Technology {jsyou, jwchung, derrick0511, park}@nlp.kaist.ac.kr Abstract Genetic information in the literature has been extensively looked into for the pur- pose of discovering the etiology of a dis- ease. As the gene-disease relation is sen- sitive to external factors, their identifica- tion is important to study a disease. En- vironmental influences, which are usu- ally called Gene-Environment interaction (GxE), have been considered as impor- tant factors and have extensively been re- searched in biology. Nevertheless, there is still a lack of systems for automatic GxE extraction from the biomedical literature due to new challenges: (1) there are no preprocessing tools and corpora for GxE, (2) expressions of GxE are often quite implicit, and (3) document-level compre- hension is usually required. We propose to overcome these challenges with neu- ral network models and show that a mod- ified sequence-to-sequence model with a static RNN decoder produces a good per- formance in GxE recognition.1 1 Introduction Identifying genetic information related to a dis- ease is an effective method for discovering the etiology of the disease. Many researchers in bi- ology have attempted to identify the relationship between different types of genetic information, such as genes, gene mutations or other biological events, and diseases. One of the difficult aspects in the research is that it is necessary to consider various external fac- tors, because they can affect whether such biolog- ical relationships hold or not. For example, it has ∗Corresponding author 1Our source code and gold standard corpus are available at http://biopathway.org/GxE been shown that there is no association of NAT2 gene and breast cancer (Zgheib et al., 2013), but after three years, other researchers made a con- flicting claim that NAT2 gene is associated with breast cancer (Kasajova et al., 2016). There may be many factors causing this difference, but inves- tigating the environmental factors has been one of the important research topics. Kasajova et al. (2016) found that NAT2 gene is associated with breast cancer when women with NAT2 gene poly- morphisms have been exposed to long-period ac- tive smoking. As a result, active smoking has been considered as a crucial factor that determines the relation between NAT2 gene and breast cancer, which biologists called gene and environment in- teraction (GxE). Since the importance of studying GxE is rec- ognized, the amount of related work has steadily been increasing (Hunter, 2005). Nonetheless, there is still a lack of systems and databases that deal with this information (Simonds et al., 2016). For the purpose of addressing this situation, we present an automatic system that extracts environ- ment terms indicating a change of gene-disease re- lations from the biomedical literature. There are three major challenges that make it difficult to perform GxE recognition using …

Content truncated to 500 words