Name
Abstract | ||
|---|---|---|
Genetic network analysis provides an important statistical strategy for the study of gene-gene interactions. Although existing methods work well in practice, several opportunities for improvement remain. For example, the regulation coefficients of some of the existing methods are not easy to solve, nor are the solutions they provide unique. Also, as genetic network analysis are typically applied to small datasets with large number of parameters, having prior knowledge about the parameters is valuable and should be incorporated into the analysis. The uniqueness of the parameter estimate and computational simplicity are also desirable in practice. To address these problems, we considered a quasi-Bayesian method for the analysis of gene regulatory networks by a multivariate linear model in which the data distribution is a quasi-likelihood, and the inference is Bayesian. This method incorporates prior information on the regulatory relationships; the set of regulation coefficients has a unique closed-form solution, and is very simple to compute. The model is evaluated by simulation and illustrated using a real dataset. This method is simple to use, permits information updating, is flexible to incorporate desired features, and has closed-form solution. Simulation studies show that the model fits the data quite well. |
Year | DOI | Venue |
|---|---|---|
2009 | 10.1142/S0219720009004059 | J. Bioinformatics and Computational Biology |
Keywords | Field | DocType |
bayesian method | Data mining,Uniqueness,Computer science,Inference,Artificial intelligence,Bioinformatics,Multivariate linear model,Gene regulatory network,Machine learning,Genetic network,Bayesian probability | Journal |
Volume | Issue | ISSN |
7 | 1 | 0219-7200 |
Citations | PageRank | References |
1 | 0.35 | 2 |
Authors | ||
3 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Ao Yuan | 1 | 3 | 3.25 |
| Guanjie Chen | 2 | 1 | 0.69 |
| Charles Rotimi | 3 | 3 | 1.15 |