Background: The primary objective of life science research is to understand complex cellular mechanisms and the interplay of various genes/proteins in multiple cellular processes. For this, PubMed is still the primary source of biomedical information even though multiple other databases such as UniProt, Protein Data Bank (PDB) and Reactome exist.
Objective: With the available large volume data from high-throughput technologies and multiple databases, finding relevant information for gene-process-phenotype has now become extremely challenging and tedious. No tool is currently available to simultaneously search PubMed and multiple other databases to get holistic information. Moreover, a typical PubMed search returns large number of articles, which need to be manually screened for identifying relevant literature. Hence, we developed BioGyan, a literature mining tool to simplify the combinatorial search for genes, celltypes and cellular processes in PubMed and other relevant databases.
Methods: BioGyan uses a robust scoring method to rank articles relevant to user search terms. The scoring method is based on the weighted sum of co-occurrence of gene, process and interactions terms in an abstract.
Results: BioGyan retrieves PubMed articles supporting association between queried genes and processes, relevant pathways from pathway databases and 3-dimensional structures from PDB. For easy viewing, all information to the user is available in single window. BioGyan showed an accuracy of 85.46% in predicting relevance of articles to a gene-process association, and performed better than PESCADOR.
Conclusion: BioGyan has several key features such as batch query of genes as well as processes, offline reading of articles, export of list of articles as bibliography and flexibility for user to revise the article relevance, making it a vital tool for literature search. Thus, BioGyan is a unique tool that offers holistic search across multiple databases while greatly automating the entire process.