Comparative Performance Analysis of the Algorithms for Detecting Periodically Expressed Genes
MetadataShow full item record
Thus far, a plethora of analysis on genome-wide gene expression microarray experiments on the cell cycle have been reported. Time series data from these experiments include gene expression profiles that might be periodically expressed. However, the numbers and actual genes that are periodically expressed have not been reported with consistency, analysis on similar experiments reports disparate numbers of genes that are periodically expressed with scant overlap. This work ultimately compares the performance of five spectral estimation schemes in their ability to recover periodically expressed genes profiles. Lomb-Scargle (LS), Capon, Missing-Data Amplitude and Phase Estimation (MAPES), Real Value Iterative Adaptive Approach (RIAA) and Lomb-Scargle Periodogram Regression (LSPR) are rigorously studied and pitted against each other in various simulated testing conditions. Results obtained using synthetic and microarray data reveals that RIAA is an efficient and robust method for the detection of periodically expressed genes in short time series data that might be characterized with noisy and irregularly sampled data points.
Agyepong, Kwadwo (2012). Comparative Performance Analysis of the Algorithms for Detecting Periodically Expressed Genes. Master's thesis, Texas A&M University. Available electronically from