Principal Component Regression Analysis for lncRNA-Disease Association Prediction Based on Pathological Stage Data

Accumulating researches have found that lncRNAs play a key role in many important biological processes, such as chromatin modification, transcription, and post-transcription regulation. Because lncRNAs play an important role in the life process, many important complex diseases have been linked to the variation and dysfunction of lncRNAs. In current prediction researches on lncRNA-disease association, clinical prognosis information of the disease (such as pathological stage, clinical stage and so on) is rarely mentioned. In this manuscript, we apply the pathological stage data into the lncRNA-disease association prediction. Firstly, coordinates reverse rotation in circular (CRRC) is proposed. 6 clusters are calculated by the proposed cluster generating algorithm (ClGeA) based on CRRC. Secondly, harmonic importance ranking (HIR) is put forward. 28 core variables are obtained by the proposed selection algorithm of core variables for cancer pathological stage (SA-CV-CPS) based on HIR and cluster. Finally, on the basis of the above 28 core variables, pathological stage prediction algorithm for lncRNA-disease association based on principal component regression analysis (PSPA-LA-PCRA) is developed. Through PSPA-LA-PCRA, principal component set (including 20 PCs) and prediction model are gained. The proposed prediction model is based on unknown human lncRNA-disease association combining with the pathological stage data. Experimental results show that better results for AUC, precision rate, recall rate and F1-score of the prediction model are achieved by PSPA-LA-PCRA, which provides a favorable research premise for subsequent prediction studies of lncRNA-disease associations.