Prediction of Human Disease Genes Based on Associations between Phenome and Proteins

Abstract

Disease genes identification is the key to the issue of human genetic diseases cure. This paper describes a new method of human genetic disease gene prediction, which is based on the relations between clinical manifestations and protein-protein interaction network. A new prediction model which is based on associated probability and Pearson correlation coefficient is also described. This mathematical model reflects the regularity of associations between similarities of phenotypes and interactions of proteins. It can discribe the real biological systems better than traditional models, and gives a stronger predictive ability.