Detection of cell-cyclic elements in mis-sampled gene expression data using a robust Capon estimator

We present a method for the estimation of possible cell-cyclic elements in mis-sampled microarray data. Accurate assessment of the frequency content of microarray data gives insight into genes which could be cell-cycle regulated. Cell-cycle regulation is one component of the complex network of genetic regulatory processes and is especially relevant to the study of cancer. As cDNA microarray experiments involve human sampling of cell populations, slight variations in the sampling times invariably occur. We propose estimating the frequency content of microarray data using the recent robust Capon estimator, and formulate a suitable uncertainty region over which to minimize. The estimator is shown to yield robust estimates with real microarray data and to identify cell-cyclic genes that elude both the traditional periodogram and the Capon spectral estimator.