| dc.contributor.author | Karakach, Tobias Kadi. | en_US |
| dc.date.accessioned | 2014-10-21T12:36:44Z | |
| dc.date.available | 2007 | |
| dc.date.issued | 2007 | en_US |
| dc.identifier.other | AAINR19598 | en_US |
| dc.identifier.uri | http://hdl.handle.net/10222/54836 | |
| dc.description | Many applications of DNA microarray technology (for gene expression monitoring), have been reported in the literature. These can be categorized as comparator-type applications, which seek to identify differential gene expression among different classes of biological samples, and serial or time course applications, which investigate temporal changes in gene expression. While data analysis methods for the former are relatively well-established, time course experiments are less common and, as a result, methods to analyze the data are not so well developed. Such experiments, nonetheless, have the potential for identifying characteristic patterns of gene expression that define the fundamental dynamics of living cells' molecular machinery, which is of great interest. In this work, a new approach to analyze time course data from spotted, two-color DNA microarrays is described and demonstrated with two data sets from the study of yeast (Saccharomyces cerevisiae) gene expression. The new method is a variation of multivariate curve resolution (MCR), which is widely used in the analysis of chemical data, and is shown to extract time profiles that are believed to be related to underlying regulatory patterns in the cell. | en_US |
| dc.description | In order to apply MCR to the microarray data sets, methods were developed to accommodate the proportional error structure and the problem of missing measurements in these data. This resulted in an algorithm referred to as multivariate curve resolution via weighted alternating least squares (MCR-WALS), whose features included freedom from assumptions about an underlying parametric model, an ability to analyze non-log-transformed data, and the use of measurement error information to obtain a weighted model that accommodates missing measurements. The first data set examined in this work is from a widely studied yeast cell cycle experiment. Profiles extracted by MCR-WALS from this data set are shown to be consistent with the expression patterns of cell-cycle-associated genes, and provide new insights into the regulation of these genes. | en_US |
| dc.description | The second data set is a yeast exit from stationary phase study and presented a greater challenge in terms of measurement quality, normalization and limited prior knowledge of the system. This necessitated the development of a new method for estimating errors associated with ratio measurements, based on a bootstrap, to deal with issues of measurement quality. In addition, a new normalization technique, referred to as sequential normalization, was introduced specifically to treat such temporary designed microarray experiments. Ultimately, the profiles extracted by MCR-WALS exhibited good reproducibility (based on a replicate experiment) and consistency with earlier results of a cluster analysis. Moreover, a cursory analysis of the gene ontology provided evidence consistent with the process of exit from stationary phase. | en_US |
| dc.description | Thesis (Ph.D.)--Dalhousie University (Canada), 2007. | en_US |
| dc.language | eng | en_US |
| dc.publisher | Dalhousie University | en_US |
| dc.publisher | | en_US |
| dc.subject | Biology, Molecular. | en_US |
| dc.subject | Chemistry, Biochemistry. | en_US |
| dc.subject | Biology, Bioinformatics. | en_US |
| dc.title | Analysis of gene expression microarray data by multivariate curve resolution. | en_US |
| dc.type | text | en_US |
| dc.contributor.degree | Ph.D. | en_US |
