Linear model tools for high-throughput gene expression data

点击翻译按钮获取中文摘要

Data generated from high-throughput platforms in plant-biological research are often analysed by linear model procedures. Checking model assumptions and subsequently transforming data or taking other measures to remedy violations poses a formidable task with these kinds of large datasets because of the need to automate the analysis pipeline as much as possible in order to save computing time and overall time needed for analysis. This project will develop procedures for critically inspecting and testing residuals from linear model fits. Moreover, data transformations and parametric link functions will be investigated regarding their suitability for highthroughput data. Particular emphasis will be given to transcriptome sequencing (RNAseq) using the Illumina sequencing technology and metabolite profiling. The proposed procedures will be investigated by simulation as well as by application to empirical datasets provided by our collaborators. A pipeline implementing these procedures will be developed and made available as an R-package. Computationally intensive components such as Monte Carlo simulation modules will be programmed in a low-level programming language.