Small Gene Networks: Finding Optimal Models Small Gene Networks: Finding Optimal Models

Genetic networks help in identifying interactions between genes, and provide information about the function role of individual genes in the cellular system. In this thesis, we have employed a Bayesian framework to learn network structures from microarray data, and implemented an algorithm that uses dynamic programming to find the optimal gene network model for a small number of genes. To test its performance, we applied the method to two distinctive synthetic datasets. ROC graphs were used to evaluate the effects of noise and a small number of samples, features that are known to be characteristic of gene expression datasets. Results showed significant improvements when the numbers of samples in a dataset were increased. The effect of adding noise to the data gave unexpected results and requires further analysis. The method was finally applied to a real microarray dataset, and led to biologically plausible results. Small Gene Networks: Finding Optimal Models