Predicting Gene-Disease Associations with Manifold Learning
In this study, we propose a manifold learning-based method for predicting disease genes by assuming that a disease and its associated genes should be consistent in some lower dimensional manifold. The 10-fold cross-validation experiments show that the area under of the receiver operating characteristic (ROC) curve (AUC) generated by our approach is 0.7452 with high-quality gene-disease associations in OMIM dataset, which is greater that of the competing method PBCF (0.5700). 9 out of top 10 predicted gene-disease associations can be supported by existing literature, which is better than the result (6 out of top 10 predicted association) of the PBCF. All these results illustrate that our method outperforms the competing method.
This work is supported in part by Natural Science and Engineering Research Council of Canada (NSERC), China Scholarship Council (CSC) and by the National Natural Science Foundation of China under Grant No. 61772552 and 61571052.
- 9.Luo, P., Tian, L.P., Ruan, J., Wu, F.X.: Disease gene prediction by integrating PPI networks, clinical RNA-Seq data and OMIM data. IEEE/ACM Trans. Comput. Biol. Bioinf. (2017, in press). https://doi.org/10.1109/TCBB.2017.2770120
- 15.Ham, J., Lee, D.D., Saul, L.K.: Semisupervised alignment of manifolds. In: AISTATS, pp. 120–127 (2005)Google Scholar