Working out drug response and large scale human disease involves study of gene to gene interactions. This involves interactions which occur between gene and transcription factors and protein protein interactions. Most of these interactions are found in Biomedical literature. There are databases like BioGRID and ChEA which are involved in curation of these interactions. However with increase in literature it becomes difficult to update these. Deep Drive is a trained system involved in extracting these interactions from literature.
In their current study Mellory et al. (2016) have build up a candidate relation with Deep dive. For each interaction they have build a probability that the gene to gene interaction is going to be correct. Their system reached 76% precision for such interactions. Their system was able to extract 3356 unique gene pairs from 100000 text articles.
Source code of the application is available at https://github.com/edoughty/deepdive_genegene_app.
Reference:
Mallory E.K. et al.(2016) Large Scale extraction of gene interaction from full text literature using Deep Dive. Bioinformatics 32(1) 106-113.