Large-Scale Gene Network Estimation

CREATED: 200809290300 Speaker: Seiya Imoto

• Model gene network as a Bayesian Network and learn the network from expression data
• Gene $g_1$: Random variable $X_1$, $x_{n,1}$ where $x_{i,j}$ is the expression value of gene j at condition i
• Selection of optimal graph by Bayes approach, can be done in $O(p2^p)$ using dynamic programming
• Constrained optimal search can improve the complexity to $O(p2^m + |S|)$, where S is the super-structure of the network