Guilt-by-association (GBA) analysis assumes that genes with similar functions will be close to each other in molecular interaction network. Many complex machine learning methods have been developed to perform GBA. In OmicsNet, we use a Random Walk With Restart (RWR) algorithm to explore the network vicinity of seed nodes and identify candidate genes that are close to them. The result will output a list of candidate genes with their associated score.
Caution: recent research has shown that GBA is not competitive with genetic association for identifying autism risk genes and the observation is likely to generalize to the study of other genetic disorders. For more details, please refer to this publication.