Biblio-MetReS for user-friendly mining of genes and biological processes in scientific documents

Anabel Usié1,2, Hiren Karathia2, Ivan Teixidó1Rui Alves2* and Francesc Solsona1*

Corresponding authors: Rui Alves This email address is being protected from spambots. You need JavaScript enabled to view it. - Francesc Solsona This email address is being protected from spambots. You need JavaScript enabled to view it.

1 Department d'Informàtica i Enginyeria Industrial, Universitat de Lleida, Av. Jaume II n°69, 25001 Lleida, Spain
2 Department de Ciències Mèdiques Bàsiques & IRBLleida, Universitat de Lleida, Montserrat Roig n°2, 25008 Lleida, Spain

PeerJ 2014, doi:10.7717/peerj.276

The electronic version of this article is the complete one and can be found online at:

© 2014 Usié et al.



One way to initiate the reconstruction of molecular circuits is by using automated text-mining techniques. Developing more efficient methods for such reconstruction is a topic of active research, and those methods are typically included by bioinformaticians in pipelines used to mine and curate large literature datasets. Nevertheless, experimental biologists have a limited number of available user-friendly tools that use text-mining for network reconstruction and require no programming skills to use. One of these tools is Biblio-MetReS. Originally, this tool permitted an on-the-fly analysis of documents contained in a number of web-based literature databases to identify co-occurrence of proteins/genes. This approach ensured results that were always up-to-date with the latest live version of the databases. However, this ‘up-to-dateness’ came at the cost of large execution times. Here we report an evolution of the application Biblio-MetReS that permits constructing co-occurrence networks for genes, GO processes, Pathways, or any combination of the three types of entities and graphically represent those entities. We show that the performance of Biblio-MetReS in identifying gene co-occurrence is as least as good as that of other comparable applications (STRING and iHOP). In addition, we also show that the identification of GO processes is on par to that reported in the latest BioCreAtIvE challenge. Finally, we also report the implementation of a new strategy that combines on-the-fly analysis of new documents with preprocessed information from documents that were encountered in previous analyses. This combination simultaneously decreases program run time and maintains ‘up-to-dateness’ of the results.