engleski

Can existing knowledge help us PREPARE in case of an epidemic crisis? Emergency exercise on Zika virus

Background: PREPARE (Platform for European Preparedness Against (Re-)emerging Epidemics) is an EU funded network for harmonised large- scale clinical research studies on infectious diseases, prepared to rapidly respond to any severe ID outbreak. PREPARE has entered Outbreak Preparation Mode in response to the current Zika virus (ZIKV) outbreak and is preparing its network to rapidly identify, diagnose and collect crucial data in case Zika arrives in Europe on a larger scale. Aim: Main aims of this exercise were: 1) to identify the steps and protocol to optimally collect and use the existing knowledge and to steer future research priorities as an initial response to a PREPARE Outbreak Response Mode ; 2) to optimise available tools for knowledge extraction, evaluation and data integration ; 3) to create ZIKV KnowledgeBase as an example with literature text mining and other publicly available data sources on the host-virus protein-protein interactions database (PPI), DrugBank-Protein interaction database, published transcriptome datasets, etc. Specific scientific aim of this exercise is to identify host factors involved in the pathogenesis of ZIKV, which could be possible drug targets, and to provide novel insights into commonalities between several neurotropic Flaviviruses, including ZIKV, Japanese encephalitis virus (JEV), Dengue virus (DENV), West Nile virus (WNV), Tick-borne encephalitis virus (TBEV) and ZIKV-specific neurodevelopmental complications. Methods: We collected full-text literature on Zika virus and abstracts on several neurotropic Flaviviruses (JEV, DENV, WNV, TBEV), with the focus on the disease/pathogen/gene/compound associations. Text mining analysis was performed to extract objects and relations from unstructured free text with the use of high precision syntax algorithm. Extracted associations were validated manually and displayed graphically into a network. We also searched for published transcriptome datasets where neuronal cells have been challenged with ZIVK or other neurotropic Flaviviruses. To compare infection induced changes in neuronal expression between neurotropic Flaviviruses, we compared the level of differentially expressed genes, checked for overrepresented pathways and overlap the results with existing knowledge from the text mining analysis and host-virus PPIs. Results: Out of total 3, 010 extracted associations, we identified 78 for ZIKV, 200 for JEV, 640 for DENV, 194 for WNV and 60 for TBEV. Only two studies on transcriptome datasets met our inclusion criteria –human microglia challenged with JEV, and human neural progenitor cells challenged with ZIKV. A total of 261 genes were found to be common to JEV and ZIKV and 177 genes in common to ZIKV and host- flavivirus PPI network at the functional level. These sets of genes may serve as prioritization of drug repurposing. Integration of text mining results with host-virus PPIs and published transcriptome datasets are currently in process to identify potential overlaps. Conclusions: KnowledgeBase is a very useful tool for emerging infections research, enabling hypotheses generation, identification of the knowledge gaps, integration of data from various sources and guidance for further research priorities setting. It may serve as a timely response tool for preparedness in case of an infectious disease epidemic crisis, within a timeframe of only several weeks. Funding: FP7 project PREPARE (grant 602525), Croatian Science Foundation (grants 8875 and 8445).

knowledge platform, Zika, text mining

nije evidentirano