aspirincode / stackppi Goto Github PK

##StackPPI

Improving protein-protein interactions prediction accuracy using XGBoost feature selection and stacked ensemble classifier.

###GcForest-PPI uses the following dependencies:

• python 3.6
• numpy
• scipy
• scikit-learn

###Guiding principles:

**The dataset file contains the S. cerevisiae, H. pylori, the independent dataset and network dataset.

**Feature extraction

1. Evolutionary information: Evolutionary_information.py is the implementation of AAC-PSSM and Bi-PSSM.
2. PseAAC.m is the implementation of PseAAC.
3. CTDC.py, CTDT.py, CTDD.py are the implementation of CTD.
4. Auto_yeast.m is the implementation of AD.

** Dimensional reduction: XGBoost.py represents XGBoost feature selection stacking_KPCA.py represents KPCA. stacking_LLE.py represents LLE. stacking_TSVD.py represents SVD. stacking_MDS.py represents MDS.

** Classifier: stacking_test.py is the implementation of the stacked ensemble classifier. yeast_Ad.py is the implementation of AdaBoost. yeast_KNN.py is the implementation of KNN. yeast_LR.py is the implementation of LR. yeast_RF.py is the implementation of RF. yeast_SVM.py is the implementation of SVM.