Abstract:
|
Interest in research vested in the discovery of biomarkers of genotypes and phenotypes continues to grow. The identification of biomarkers could allow for several potential benefits such as disease diagnostic capabilities utilizing readily accessible biofluids rather than more invasive procedures. However, traditional statistical methods can become inadequate or inappropriate for data available for biomarker discovery, generally due to the scale and complexity of the data. We propose and demonstrate a statistical meta-analysis approach on a large-scale population proteomics dataset with large batch-effects. In addition to appropriately dealing with batch-effects, we demonstrate that the meta-analysis methodology aids in alleviating issues with missing data, non-convergence in model estimation, and violation of model assumptions. Additionally, data at the protein level further validates findings made at the peptide level.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.