Testing dependence/correlation of two variables is one of the fundamental tasks in statistics. In this work, we proposed a powerful method for nonlinear dependence of two continuous variables (X and Y).
We addressed this research question by using BNNPT (Bagging Nearest-Neighbor Prediction Test, software available at https://sourceforge.net/projects/bnnpt/). In the BNNPT framework, we first used the value of X to construct a bagging neighborhood structure. And then, we got the out of bag estimator of Y based on the bagging neighborhood structure. The square error was calculated to measure how good Y is predicted by X. Finally, permutation test was applied to detect the significance of the observed square error.
To evaluate the strength of BNNPT compared to seven other methods, we performed extensive simulations to explore the relationship between methods and compared the false positive rates and statistical power using both simulated and real RNA-seq datasets.
Features
- Nonlinear correlation/dependence
- bagging nearest neighbor
- powerful method
- permutation test
- real kidney RNA-seq datasets