Associated Disease Frequency-based Measure for Finding Candidate Target Genes from the Biomedical Literature

Abstract

The identification of the side-effects of chemicals is the serious and costly stage in the drug development. Most side-effects are caused by their toxicity and also their correlation with other diseases than target diseases. We present a novel measure that identifies disease-associated genes from the biomedical literature in terms of causing less side-effects. This enables the identification of specific disease-associated genes, the decreased expression of which would result in a lower probability of side-effects, thus contributing to efficient drug development. Our method evaluates the specificity of a gene to a particular disease based on the number of associated diseases with the gene. In addition, we consider transitive gene-disease associations, that is, indirect gene-disease associations via intermediate genes. Gene-disease associations are extracted from the PubMed abstracts based on term co-occurrences. Also, we discuss the ranking results for Alzheimer disease and various cancers to verify the effectiveness of our measure. Ranking results for other diseases are available at http://www.ps.noda.tus.ac.jp/ddss/.