Abstract
The scientific literature contains large amounts of information on genes, proteins, chemicals and their interactions. Extraction and integration of this information in curated knowledge bases help researchers support their experimental results, leading to new hypotheses and discoveries. This is especially relevant for precision medicine, which aims to understand the individual variability across patient groups in order to select the most appropriate treatments. Methods for improved retrieval and automatic relation extraction from biomedical literature are therefore required for collecting structured information from the growing number of published works. In this paper, we follow a deep learning approach for extracting mentions of chemical–protein interactions from biomedical articles, based on various enhancements over our participation in the BioCreative VI CHEMPROT task. A significant aspect of our best method is the use of a simple deep learning model together with a very narrow representation of the relation instances, using only up to 10 words from the shortest dependency path and the respective dependency edges. Bidirectional long short-term memory recurrent networks or convolutional neural networks are used to build the deep learning models. We report the results of several experiments and show that our best model is competitive with more complex sentence representations or network structures, achieving an F1-score of 0.6306 on the test set. The source code of our work, along with detailed statistics, is publicly available.
Highlights
As the knowledge of how biological systems work at different structural levels grows, more possibilities arise for applying it in diagnosing and treating common and complex diseases
As noted in the previous section, the use of different random states generates different training and validation subsets, which in turn results in different trained models. This approach allows using a large amount of data for early stopping, which in our preliminary experiments proved important for improving generalization, while still using most of the available data for training
The three best results on the development set (F1-scores: 0.6496, 0.6473 and 0.6385) were obtained by the bidirectional long short-term memory (BiLSTM) model using only the shortest dependency path (SDP) with word and dependency features where different embedding models are used, being the highest result achieved with the biomedical word embeddings created by Chen et al [57]
Summary
As the knowledge of how biological systems work at different structural levels grows, more possibilities arise for applying it in diagnosing and treating common and complex diseases. Relevant fine-grained information is constantly being communicated in the form of natural language through scientific publications To exploit this source of updated knowledge, several methods have been proposed for retrieving relevant articles for database curation [2], and for extracting from the unstructured texts information such as entity mentions [3, 4], biomolecular interactions and events [5, 6] or the clinical and pharmacological impact of genetic mutations [7]. These methods have proven essential for collecting the most recent research results and for expediting database curation [8]. The development of systems able to automatically extract such relations may expedite curation work and contribute to the amount of information available in structured annotation databases, in a form that is searched and retrieved by researchers
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Database : the journal of biological databases and curation
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
