Genomics and natural language processing

Today, the computational exploration and management of large text repositories are usually accomplished with search engines and databases that are based on a suite of text processing, indexing and search tools that are referred to collectively as 'natural language processing' (NLP) technologies. There are three fundamental aspects to NLP: information retrieval, semantics and information extraction. Exploring and managing the biomedical literature with these technologies, however, presents some interesting challenges, primarily because of the relationships between biomedical texts and biological sequences. The associations between biological sequences and texts are a truly unique aspect of the biomedical literature. However, understanding the complex associations that exist between genes, sequences and texts is a daunting task. The flood of sequence information produced by the rapid advances in genomics is creating new ways to explore texts and is blurring the traditional lines that separate bioinformatics and NLP. Biological NLP (bio-NLP) is an emerging field of research that seeks to create tools and methodologies for sequence and textual analysis that combine bioinformatics and NLP technologies in a synergistic fashion. Some bio-NLP researchers are focusing on texts as a means to discover information about protein interactions, and are wrestling with how best to adapt traditional NLP technologies to this task. Others, taking a more sequence-centred approach, are exploring the use of texts as a means to improve sequence-retrieval algorithms and as an aid to sequence annotation. If bio-NLP is to achieve its full potential, it will have to move beyond information management and generate specific predictions pertaining to gene function that can be verified at the bench. The synergistic use of sequence and text to extract latent information from the biomedical literature holds much promise in this regard. Realizing this potential, however, will require more and better ontologies, software that is able to make inferences using sequence and textual information, and access to the full text of articles.