Abstract:
|
Combining information from multiple experiments that address the same or correlated hypothesis can provide meaningful answers to important biological questions. We restrict our attention to situations in which the multiple experiments aim to determine common genomic variants across multiple tissue types. For example, our cancer biomarker project entails profiling several biological matrices (e.g., tissue, plasma, serum, and urine) from the same subjects. Meta-analysis techniques and various Bayesian approaches have been used to combine results from multiple independent studies analyzing the same matrix and some work has been done combining different types of omics data (e.g., proteomics and genomics studies). However, those methods have limitations related to the underlying assumption of independency on measurements between experiments and thus require modification to reflect the dependent relationships among datasets from multiple experiments. Herein, we examine a multivariate permutation approach to account for the intra-subject correlation between different matrices. We compare this method with other multivariate tests using simulations with different data configurations.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.