A fundamental problem in genome-wide studies is the detection of differential expression among thousands of gene expression values produced by micro array technology. Although the problem is most naturally expressed as a sequence of hypothesis tests involving the expression distributions of individual genes, two points have recently been noted: 1) there is a strong dependence between expression levels of different genes that may (at least partially) be attributable to their involvement in various pathways, and 2) information about which sets of genes form pathways has become more readily available. This has led to interest in the problem of the testing of multivariate gene expressions associated with known pathways. While adding potential information, this idea complicates the application of hypothesis testing in that multivariate testing is inherently more challenging than univariate testing, and requires a much more thorough evaluation of available procedures. Given the complications inherent in multidimensional testing, a rigorous development of a suite of hypothesis tests for pathway specific genes can be expected to result in increased detection of experimental effects in the broadest sense. We further anticipate that the development of pathway hypothesis tests may eliminate the need in some applications to compare an experimental condition to a control, provided sufficiently rich pathway data is available. The resulting methodology will be made available for general use through the development of appropriate software applications.
The proposed methods of data analysis are forecast to yield a better understanding of pathway organization and gene-gene relationships. Therefore, they have the potential to drive a disruptive understanding of disease mechanisms and treatments.
| Almudevar, Anthony; LaCombe, Jason (2012) On the choice of prior density for the Bayesian analysis of pedigree structure. Theor Popul Biol 81:131-43 |
| McCall, Matthew N; Almudevar, Anthony (2012) Affymetrix GeneChip microarray preprocessing for multivariate analyses. Brief Bioinform 13:536-46 |
| Needham, Mark; Hu, Rui; Dwarkadas, Sandhya et al. (2011) Hierarchical parallelization of gene differential association analysis. BMC Bioinformatics 12:374 |
| Glazko, Galina V; Emmert-Streib, Frank (2009) Unite and conquer: univariate and multivariate approaches for finding differentially expressed gene sets. Bioinformatics 25:2348-54 |
| Almudevar, Anthony (2009) A note on the calculation of N-statistics. J Bioinform Comput Biol 7:895-903 |
