TY - JOUR TI - Robust gene set analysis and robust gene expression DO - https://doi.org/doi:10.7282/T31J982J PY - 2014 AB - This paper explores various methods of statistical analysis of DNA microarray data. First, we review the RMA method which produces estimates of gene expression from a microarray data and propose a new version of RMA that is not only resistant to outliers but also has high efficiency. To construct our new RMA estimator we rely upon M-estimator of location, including Tukey’s biweight and Huber’s M-estimator. We compare the performance of our robust version of RMA with median, the currently used one in the RMA method, as well as mean, which is a non-robust estimator of location. Second, we review the Gene Set Enrichment Analysis (GSEA) methodology. Currently, the GSEA method is performed at gene-level. This requires DNA microarray data be transformed from the raw probe-level data to the gene-level data. This process cannot avoid losing subtle but crucial information contained in the probe-level data. Inspired by the GSEA method, we extend its idea to the probe-level data. Finally, we develop a family of enrichment method - Enrichment Analysis using M-estimator (EAME), which, as implied by its name, uses robust M-estimator and take advantage of the idea of gene set enrichment. At the end of this paper, we use the R language as a tool to show some examples of DNA microarray analysis based on the methodologies discussed in this paper. KW - Statistics and Biostatistics KW - DNA microarrays KW - Gene expression LA - eng ER -