Fig. 1.

The Pipeline of the LPI-HyADBS framework. (1) Initial feature acquisition. lncRNA and protein features are acquired with Pyfeat [51] and concatenated to depict each lncRNA-protein pair. (2) Feature selection. The concatenated features are reduced based on AdaBoost. (3) LPI classification. DNN, XGBoost, and C-SVM are designed to capture unobserved LPIs, respectively. (4) Ensemble. An ensemble framework is proposed to combine prediction results from the three classifiers