Prediction of Disease-associated Genes by advanced Random Walk with Restart on Multiplex and Heterogeneous Biological Networks
Résumé
Rare monogenic diseases globally affect millions of persons, but many causative genes remain to be discovered. Several computational approaches have been developed to predict disease-associated genes. Guilt-by-association strategies on protein interaction networks, in particular, postulate that proteins lying in a close network vicinity are functionally-related and implicated in similar phenotypes. However, current network approaches are limited as they do not exploit the richness of biological networks, which are both multiplex (i.e., containing different layers of physical and functional interactions between genes and proteins), and heterogeneous (i.e., containing both interactions between genes/proteins, and interactions between diseases). In the present study, we extended the Random Walk with Restart algorithm to leverage these complex biological networks. We compared our algorithm to classical random walks thanks to a leave-one-out strategy. The Random Walk with Restart on multiplex and heterogeneous networks takes advantage of data pluralism and shows increased performances to predict known disease-associated genes. We finally applied it to predict candidate genes for the Wiedemann-Rautenstrauch Syndrome.
Origine | Fichiers éditeurs autorisés sur une archive ouverte |
---|