Genomic data often comes in a form which is noisy and sparse. It is challenging to recover the truth due to its complex structure and high dimensionality.
We propose to use Nonparametric Empirical Bayesian schemes to tackle the problem. The method adapts especially well to varying degrees of sparsity. It not only performs well to recover the signals, but also provides credible intervals. We also propose a method to control FDR in the case of multiple testing.
Joint work with J. Cai and Y. Ritov