Figure 1
From: A fast and high performance multiple data integration algorithm for identifying human disease genes
![Figure 1](http://media.springernature.com/full/springer-static/image/art%3A10.1186%2F1755-8794-8-S3-S2/MediaObjects/12920_2015_Article_546_Fig1_HTML.jpg)
The general idea of the proposed logistic regression based algorithm. (a) A prior probability of each gene is first predefined. (b) The class label of each gene is then assigned according to its prior probability. (c) A biological network gives the neighborhood connections between individual genes. (d) A feature matrix is constructed based on the labels of individual vertices and the biological network. (e) A binary logistic regression is conducted by using class labels as categorical dependent variables and individual features as predictor variables. (f) A posterior probability is obtained from the binary logistic regression for each unknown genes. (g) The posterior probability is transformed into a decision score for each unknown genes. (a) - (f) make up the inference stage, while (g) is the decision stage.