Degree-adjusted algorithm for prioritisation of candidate disease genes from gene expression and protein interactome

Computational methods play an important role in the disease genes prioritisation by integrating many kinds of data sources such as gene expression, functional annotations and protein-protein interactions. However, the existing methods usually perform well in predicting highly linked genes, whereas they work quite poorly for loosely linked genes. Motivated by this observation, a degree-adjusted strategy is applied to improve the algorithm that was proposed earlier for the prediction of disease genes from gene expression and protein interactions. The authors also showed that the modified method is good at identifying loosely linked disease genes and the overall performance gets enhanced accordingly. This study suggests the importance of statistically adjusting the degree distribution bias in the background network for network-based modelling of complex diseases.