Grace-AKO: a novel and stable knockoff filter for variable selection incorporating gene network structures
Published in BMC Bioinformatics, 2022
Grace-AKO is a high-dimensional variable selection method that combines graph-constrained estimation with aggregated knockoff inference. By incorporating prior gene network structures, the method improves selection stability while maintaining finite-sample false discovery rate control.
Recommended citation: Tian, P., Hu, Y., Liu, Z., Zhang, Y. D. (2022). "Grace-AKO: a novel and stable knockoff filter for variable selection incorporating gene network structures." BMC Bioinformatics, 23, 478.
Download Paper