Ontology based text mining of gene-phenotype associations: application to candidate gene prediction

Gene-phenotype associations play an important role in understanding
  the disease mechanisms which is a requirement for treatment
  development. A portion of gene-phenotype associations are observed
  mainly experimentally and made publicly available through several
  standard resources such as MGI. However, there is still a vast
  amount of gene--phenotype associations buried in the biomedical
  literature. Given the large amount of literature data, we need
  automated text mining tools to alleviate the burden in manual
  curation of gene-phenotype associations and to develop
  comprehensive resources. We developed an ontology based
  approach in combination with statistical methods to text mine
  gene-phenotype associations from literature. Our method achieved
  AUC values of 0.90 and 0.75 in recovering known gene-phenotype
  associations from HPO and MGI respectively. We posit that candidate
  genes and their relevant diseases should be expressed with similar
  phenotypes in publications. Thus, we demonstrate the utility of our
  approach by predicting disease candidate genes based on the semantic
  similarities of phenotypes associated with genes and diseases. We evaluated our disease candidate prediction model on
  the gene-disease associations from MGI. Our model achieved AUC
  values of 0.90 and 0.87 on OMIM (human) and MGI (mouse) datasets of
  gene-disease associations respectively. Our manual analysis on the
  text mined data revealed that, our method can accurately extract
  gene-phenotype associations which are not currently covered by the
  existing public gene-phenotype resources. Overall, results indicate
  that our method can precisely extract known as well as new
  gene-phenotype associations from literature. This released dataset at Zenodo covers our gene-phenotype extracts from the literature. All the methods used to extract the data are available at https://github.com/bio-ontology-research-group/genepheno.