Skip to main content
. 2019 Dec 24;20(Suppl 25):700. doi: 10.1186/s12859-019-3275-6

Fig. 4.

Fig. 4

The flow of the stacking algorithm. Model_1 refers to the sequence information model, and Model_2 refers to the evolutionary information model. The class probability outputs of two models are combined to serve as a new training set to train the meta-classifier (MLR). When testing unknown membrane protein samples, we take the result of the meta-classifier as the final output