PAN - Posterior association networks
The package implements posterior association networks (PANs) to encode posteriori beliefs of functional association types on edges and phenotypic information on nodes. Instead of making an arbitrary cutoff on correlation coeffcients, PAN quantifies the statistical significances of edges by performing a beta-mixture modelling of gene association densities.
A brief overview of PAN:
- Quantifying gene associations: PAN prefers the uncentered correlation coefficient (also known as cosine similarity), which was described as a function that considers both magnitude and direction. It has been proved to be a highly desirable metric for exploring gene expression patterns
- Beta-mixture modelling: PAN conducts beta-mixture modelling on association densities to quantify the significances of gene interactions. Prior knowledge of functional associations such as protein-protein interactions can also be incorporated.
- Inferring a posterior association network: To infer a PAN, we compute posterior odds for each gene pair in favor of signal (positive or negative association) to noise (dissociation). These posterior odds can be used to weigh edges of a PAN, and a cutoff score (e.g. 10) can be selected to exclude non-significant edges.
- Searching for enriched functional modules: PAN performs hierarchical clustering on second-order similarities, a popular measure of gene modularity, between genes to search for enriched modules. To assess the uncertainty of the clustering analysis, PAN computes a p-value for each cluster using multiscale bootstrap resampling powered by pvclust.
Download and Installation
Posterior Association Networks are implemented in R-package PANR available from Bioconductor.
To install this package, start R and enter:
And for the data package type:
Posterior association networks and functional modules inferred from rich phenotypes of gene perturbations
X. Wang, M.A. Castro, K.W. Mulder^, F. Markowetz^
PLoS Comp Bio 8(6): e1002566
PMID:22761558 | doi:10.1371/journal.pcbi.1002566
Diverse epigenetic strategies interact to control epidermal differentiation
K.W. Mulder, X. Wang, C. Escriu, Y. Ito, R.F. Schwarz, J. Gillis, G. Sirokmany, G. Donati, S. Uribe-Lewis, P. Pavlidis, A. Murrell, F. Markowetz, F. Watt
Nature Cell Biology 14(7), 753-763 (2012)
PMID:22729083 | doi:10.1038/ncb2520