Text Mining for Translational Bioinformatics

Hong-Jie Dai, Chih-Hsuan Wei, Hung-Yu Kao, Rey-Long Liu, Richard Tzong-Han Tsai, Zhiyong Lu
2015 BioMed Research International  
Translational bioinformatics is an emerging field with a fascinating aim to develop novel computational techniques to facilitate traditional translational research through the convergence of molecular bioinformatics, biostatistics, statistical genetics, and clinical informatics. Translational bioinformatics is now more powerful than ever and has become a clinch between biological findings and clinical informatics. The computational techniques contribute by integrating multidimensional data
more » ... sting of medications, diseases, and genomes with clinical and pathological features. They are applied in various aspects with the hope of uncovering therapeutic targets and biomarkers of patient response. However, the accumulation of rich data from past studies, advancement of new experimental techniques, and ease of access to publications nowadays result in enormous repositories of scientific literatures and biomedical data, hindering the translation of molecular understandings into technologies that could impact patients. Text mining is an established field, but its application for translational bioinformatics is still a novel research direction with enormous research potential. The present issue emphasizes the application of text mining on biomedical/clinical publications and knowledge bases to facilitate the discovery and management of translational medical research knowledge. Rapid growth of disease related biomedical literature makes the traditional information retrieval techniques insufficient to fulfill searchers' information needs. In the paper "Disease Related Knowledge Summarization Based on Deep Graph Search," X. Wu et al. developed an approach which is able to automatically retrieve disease related knowledge in a summarized form from the large volume of online biomedical literature. This approach is capable of finding both direct relations between diseases and genes as well as indirect obscure relationships among diseases and other biomedical entities. Their experiment results show that a precision of 0.6 and a recall of 0.61 can be achieved on extracting bladder cancer-related genetic entities compared to a reference standard recorded in the Online Mendelian Inheritance in Man (OMIM) and Genetics Home Reference database. The large amount of biomedical literature provides useful knowledge resource for researchers to form biomedical hypotheses. In their work entitled "Supervised Learning Based Hypothesis Generation from Biomedical Literature," S. Sang et al. proposed a supervised learning-based approach to generate hypotheses from biomedical literature. This approach splits the traditional processing of hypothesis generation model into two models, the AB model and the BC model, which are constructed with a supervised learning method. The purpose of the AB model is to determine whether a physiological phenomenon is caused by a disease in a sentence, and the BC model is used to judge whether there exists an entity having physiological effects on human beings in a sentence. The experimental results on the three classic Swanson's hypotheses demonstrate that the proposed approach can achieve better performance in comparison
doi:10.1155/2015/368264 pmid:26380272 pmcid:PMC4563058 fatcat:mshc5klqsrb2vd7ik7jcauazey