Statistical methodologies for the analysis and normalization of RIP-Chip data
Year: 2011
Type: dissertation
Abstract: Pre-mRNA splicing is an essential step in the post-transcriptional gene expression control involving protein-splicing factors like PTB and U2AF65; the last one is exported to the cytoplasm and involved in some other cellular functions. The identification of PTB- and U2AF65-associated mRNAs under native conditions was performed by immunoprecipitation and hybridization on Chip (RIP-Chip) technology using the Affymetrix GeneChip R Human Genome U133 Plus 2.0. The aim of this thesis is to develop statistical methodologies for low level analysis and enriched gene selection in RIP-Chip experiments. When the most common methodologies for quality assessment, low level analysis (background adjustment, normalization and summarization) and detection of diferentially expressed genes (DEG), are applied to RIP-Chip data the obtained results difer. This probably happens because usually more than 20% of the mRNAs are enriched, while methods for normalization and identification of DEG are developed supposing that only a small proportion of genes (1% or 5%, say) express diferently. Also, methods for detecting diferentially expressed genes may not be the most adequate for gene enrichment selection. In this thesis is implemented a background correction method inspired in a non-specific hybridization method used for pre-processing ChIP-Chip data. Linear regression models are used in each array to model the non-specific hybridization. Probe intensities on the array are standardized using their predicted intensity and the variance of similar predicted intensities. The standardized probe intensities showed no need for further normalization, so the scores could be directly compared. It is proposed a probe set score, a probe set enrichment value and its p-value for enriched gene selection. The genes selected using this new method are practically the same as the ones found experimentally. Additionally, a new methodology based on ranks is presented for enriched gene selection, being applied to the probe set scores proposed. Both methodologies had high accuracy when applied to Spike-In U133 dataset, which is used to benchmark methodologies for analysing Affymetrix microarrays.
Author: Emiliano Barreto Hernandéz
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W2262807102