20180503 UWC NGS C Gene Expression Microarray RNA Seq

Показать описание

This year I contribute three lectures for the Next-Generation Sequencing module, taught for the B.Sc. Honours program at the University of the Western Cape Department of Biotechnology. This, the third lecture, addresses the topic of gene expression technologies and bioinformatics. I began by comparing two dominant platforms for measuring the transcriptome: the microarray and the "RNA-Seq" cDNA sequencer. We discussed the differences in their detection methods. The microarray measures the amount of cellular mRNA (via cDNA) that hybridizes to DNA probes that have been affixed to or synthesized on a microarray and then measures the amount of cDNA present through fluorescence. RNA-Seq produces read counts for fragments of cDNA that have been randomly sampled throughout their sequences. Though analyzing RNA-Seq data is much more complex, a variety of applications are possible in RNA-Seq that are considerably less feasible from microarrays: novel SNV detection, gene fusion discovery, high dynamic range measurements, and measurements in non-model organisms. We discussed the range of steps that we find in common between the methods, as well: normalization to avoid systematic variation, the detection of abundance-based fold-change bias through MA Plots, the use of clustering to recognize sample-to-sample similarity and gene-to-gene similarity, the representation of p-values and log fold-changes in volcano plots, and the use of machine learning to develop phenotypic classifiers based on gene expression. From there, we shifted to important topics in biostatistics, such as the difference between T-tests and U-tests (magnitudes versus rank orders) and the importance of multiple testing correction. We examined Bonferroni Adjustment and Benjamini-Hochberg FDR approaches in a closer look.