Infrastructure for Semantic Annotation in the Genomics Domain

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 6921–6929 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 6921 Infrastructure for Semantic Annotation in the Genomics Domain Mahmoud El-Haj1 Nathan Rutherford1 Matt Coole1 Ignatius Ezeani1 Sheryl Prentice1 Nancy Ide2 Jo Knight1 Scott Piao1 John Mariani1 Paul Rayson1 and Keith Suderman3 1 Lancaster University, UK 2, 3 Vassar College, USA 1 {m.el-haj, n.rutherford, m.coole, i.ezeani, s.r.prentice1, jo.knight, s.piao, j.mariani, p.rayson}@lancaster.ac.uk 2 ide@vassar.edu, 3 suderman@cs.vassar.edu Abstract We describe a novel super-infrastructure for biomedical text mining which incorporates an end-to-end pipeline for the collection, annotation, storage, retrieval and analysis of biomedical and life sciences literature, combining NLP and corpus linguistics methods. The infrastructure permits extreme-scale research on the open access PubMed Central archive. It combines an updatable Gene Ontology Semantic Tagger (GOST) for entity identification and semantic markup in the literature, with a NLP pipeline scheduler (Buster) to collect and process the corpus, and a bespoke columnar corpus database (LexiDB) for indexing. The corpus database is distributed to permit fast indexing, and provides a simple web front-end with corpus linguistics methods for sub-corpus comparison and retrieval. GOST is also connected as a service in the Language Application (LAPPS) Grid, in which context it is interoperable with other NLP tools and data in the Grid and can be combined with them in more complex workflows. In a literature based discovery setting, we have created an annotated corpus of 9,776 papers with 5,481,543 words. Keywords: BioNLP, Ontology, Semantic Tagger, Corpus, PubMed, Genomics, Infrastructure 1. Introduction In many fields, academics rely only on full text searching and citation networks to find related research. In the medical domain, much research has been undertaken in literature- based discovery that relies on knowledge and information extraction techniques to perform automated hypothesis gen- eration, in order to find new relationships between existing knowledge. High-level semantic taxonomies and networks can be applied to achieve broad linking and identification of semantic categories, but these fail to identify and dis- ambiguate sub-discipline-specific terminology, or indeed to cope with the continued expansion and development of do- main specific terminologies. For example, a search in the main biomedical literature citation database (PubMed) for the term ‘genome wide association study’ results in just five papers from 1995, 141 from 2005 and 3,633 from 2015. At the same time, the domain terminology in genomics has developed, expanded and changed, meaning that broad cov- erage semantic taxonomies cannot keep in step. In this paper, we propose a novel combination of Nat- ural Language Processing (NLP) and Corpus Linguistics (CL) methods and tools, connected together via SOAP and REST API calls in a loosely-coupled open infrastructure intended to provide a range of facilities for researchers to collect, annotate, store and retrieve large corpora derived from open access biomedical and life sciences literature. All the tools used were developed by the authors of this paper. NLP annotation tools draw from an existing Gene Ontology (GO) which is updated monthly, and facilitate …

Content truncated to 500 words