Integrating Large-Scale Text Mining and Co-Expression Networks: Targeting NADP(H) Metabolism in E. coli with Event Extraction

Abstract

We present an application of EVEX, a literature-scale event extraction resource, in the concrete biological use case of NADP(H)<br />metabolism regulation in Escherichia coli. We make extensive use of the EVEX event generalization based on gene family definitions<br />in Ensembl Genomes, to extract cross-species candidate regulators. We manually evaluate the resulting network so as to only preserve<br />correct events and facilitate its integration with microarray-based co-expression data. When analysing the combined network obtained<br />from text mining and co-expression, we identify 41 candidate genes involved in triangular patterns involving both subnetworks. Several<br />of these candidates are of particular interest, and we discuss their biological relevance further. This study is the first to present a<br />real-world evaluation of the EVEX resource in particular and literature-scale application of the systems emerging from the BioNLP<br />Shared Task series in general. We summarize the lessons learned from this use case in order to focus future development of EVEX and<br />similar literature-scale resources.