Skip to main content
. 2016 Dec 21;11(12):e0168392. doi: 10.1371/journal.pone.0168392

Fig 1. Infrastructure of the meta-predictor.

Fig 1

Query sequence is input into each individual predictor. The outputs of individual predictors are preprocessed and then fed into an ANN to make a new prediction, which is the output of meta-predictor. Therefore, the meta-predictor is composed of individual predictors, preprocessing modules, and ANN. The parameters of ANN will be trained using datasets containing both positive and negative samples of miRNAs. Although five individual predictors were shown in the figure, the meta-predictor could be made from any number of individual predictors out of five. The total number of possible meta-predictors is 26. The final meta-predictor mirMeta contains all five individual predictors.