Alignment algorithms for gene expression time series data

The sampling time or rate of progress of gene expression time series belonging to a biological process can vary in different conditions or in individuals. Therefore, the series need to be aligned with alignment algorithms special to gene expression time course data before comparison, classification and clustering tasks. In this study, it is aimed to present a general overview of the different approaches used in aligning gene expression time series data, and to apply a commonly used algorithm by using human and mouse fetal lung development data. First of all, gene selection criteria before the alignment and, mathematical and statistical modeling approaches for handling missing data, irregular sampling and sparseness of gene expression time series data problems have been discussed. Then, the use of dynamic time warping algorithm that is widely used in different research areas such as speech recognition, data mining and information retrieval to aligning gene expression data has been examined. After discussion about dynamic time warping algorithm other developed methods such as parametric time warping, multi segment generative time warping and shorting correlation optimized time warping algorithms have been summarized with their modeling approaches. B-splines, its coefficients estimated via linear mixed effect model, have been constructed for the time series in pre-processing step of parametric time warping; and smoothing B-splines have been used for the last two methods. Finally, the performance of these alignment algorithms, their effectiveness according to application areas, and their limitations were evaluated. In application section of this thesis, time shifts between mouse and human fetal lung development have been estimated with dynamic time warping algorithm, and statistical significance of these shifts has been determined via a simulation procedure. The algorithms used to align the gene expression time series provide great ease in bring out the similarities and differences in the biological events in different speeds at the genome level.