Towards Structuring Unstructured GenBank Metadata for Enhancing Comparative Biological Studies

Within large sequence repositories such as GenBank there is a wealth of metadata providing contextual information that may enhance search and retrieval of relevant sequences for a range of subsequent analyses. One challenge is the use of free-text in these metadata fields where approaches are needed to extract, structure, and encode essential information. The goal of the present study was to explore the feasibility of using a combination of existing resources for annotating unstructured GenBank metadata, initially focusing on the "host" and "isolation_source" fields. This paper summarizes early results for 10 host organisms that include a characterization of associated isolation sources with respect to biomedical ontologies and semantic types. The findings from this preliminary study provide insights to the rich amount of information captured within these unstructured metadata, guidance for addressing the challenges and issues encountered, and highlight the potential value for enriching comparative biological studies towards improving human health.