A Study Of Similarity Between Diseases And Similarity Between Genes Based On Disease Ontology

The construction of Disease Ontology (DO) promotes the investigation ofdiseases and disease risk factors. It enables us to analyze the disease similarity byadopting semantic similarity measures, which helps us to study the relationship ofdifferent diseases and classify diseases. Simultaneously, similarity between genes canalso be analyzed by their associations with similar diseases, which helps us tounderstand disease heterogeneity and gain insights into molecular pathogenesis ofsimilar diseases. Hence, we developed an R-based software package (DOSim) tomeasure the similarity between diseases, measure the similarity between human genesin terms of diseases, and conduct other utilities for performing DO enrichmentanalysis, detecting disease-driven gene modules, annotating the modules on diseases,functions and pathways, and visualizing DO structures, etc. DOSim can reflect themodular characteristics of disease related genes and promote our understanding of thecomplex pathogenesis of diseases. DOSim focuses on the computation of diseasesimilarity and gene similarity in terms of diseases. An application of128differentcancers by applying disease similarity demonstrates the relationship between cancers.The further hierarchical clustering of these128different cancers shows modularcharacteristics and the diseases in the same module presents hierarchical structure inthe DO graph. Another application of361obesity-related genes by applying genesimilarity shows the complex pathogenesis of obesity. In addition, the gene moduledetection and gene module multilayered annotation on these361obesity-related genesin DOSim promote our understanding of the complex pathogenesis of obesity riskphenotypes and the heterogeneity of obesity-related diseases. DOSim is available onthe CRAN or http://bioinfo.hrbmu.edu.cn/dosim...