Spectrum estimation is a popular method for identifying periodically expressed genes in microarray time series analysis.For unevenly sampled data, a common technique is applying the Lomb-Scargle algorithm. The performance of this method suffers from the effect of noise in the data. In this paper, we propose a new spectrum estimation algorithm for unevenly sampled data. The new method is based on signal reconstructing technic in aliased shift-invariant signal spaces and a direct spectrum estimation formula was derived based on B-spline basis. The new algorithm is very flexible and can reduce the effect of noise by adjusting the order of B-spline basis. The test on simulated noisy signal and typical periodically expressed gene data shows our algorithm is accurate compared with Lomb-Scargle algorithm.
Spectrum estimation eriodically expressed gene unevenly sampled data Lomb-Scargle algorithm signal reconstruction B-spline
JUN XIAN SHUAN-HU WU ALAN LIEW DAVID SMITH HONG YAN
Department of Electronic Engineering, City University of Hong Kong, Hong Kong;Department of Mathemat Department of Electronic Engineering, City University of Hong Kong, Hong Kong;School of Computer Sci Department of Computer Science and Engineering Chinese University of Hong Kong, Shatin, Hong Kong Department of Biochemistry, University of Hong Kong, Pok Fu Lam, Hong Kong Department of Electronic Engineering, City University of Hong Kong, Hong Kong;School of Electronic a