Identifying periodicity of microarray gene expression profiles by autoregressive modeling and spectral estimation

We proposed an effective algorithm to analyze the periodicity of noisy microarray time series data. Each DNA microarray chip produces thousands of gene expressions. The data have a high level of noise, which make it a challenge to identity characteristics of the genes. Our algorithm is based on singular value decomposition (SVD), singular spectrum analysis (SSA) and autoregressive (AR) model-based spectral estimation. We have applied our algorithm to simulated noisy gene expression profiles and Plasmodium falciparum data, and the result shows that the algorithm is able to remove noise such that periodic genes expression profiles can be identified accurately.