Projects

• Robust intervention in gene regulatory networks (TGen)

Sponsors: NSF, W. M. Keck Foundation
Description: This project involves the design of intervention strategies for altering the dynamics of gene regulatory networks that are robust with respect to modeling uncertainties.

• Approximation of optimal control in gene regulatory network (TGen)

Sponsors: NSF, W. M. Keck Foundation
Description: To ease the curse of dimensionality, this project investigates approximation approaches to obtain sub-optimal control policies in the application of control to gene networks.

• Application of optimal control to cancer therapy(TGen)

Sponsors: NSF
Description: The study applies control theory to find optimal treatment policies for cancer in the context of gene regulatory networks.

• Inference of discrete gene regulatory networks(TGen, University of Texas at San Antonio, Strathclyde University)

Sponsors: NSF, NCI
Description: This effort consists of a number of approaches for designing generegulatory networks, including using information theory, multiple datasources, and prior information in a Bayesian framework.

• Inference of continuous-time gene regulatory networks(Texas A&M at Praireview)

Sponsors: NSF
Description: This project aims to construct continuous-time gene networks from time-course data.

• Characterization of multivariate gene prediction(University of Sao Paulo, TGen)

Sponsors: University of Sao Paulo
Description: The study investigates the probabilistic characteristics of genes that lack predictive power for the system when observed alone but possess high predictive power when observed in combinations

• Classifier error estimation(TGen)

Sponsors: NSF, NCI, University of Paris
Description: This effort involves several research thrusts, including studying the joint distribution between the true and estimated errors, and the design of new bolstering and convex estimators.

• Feature selection with small samples(TGen)

Sponsors: TGen
Description: This study investigates the small-sample performance of feature-selection algorithms, such as their ability to find good feature sets and their susceptibility to peaking.

• Context analysis for classification(TGen)

Sponsors: University of Paris, TGen
Description: This project analyzes the effects of mixed populations and proposes to find sub-samples which represent characterized populations, or contexts, within which classification is feasible.

• Microarray compression (TGen)

Sponsors: NSF
Description: This study considers the effect of microarray image compression on phenotype classification and the detection of significant up- or down-regulated genes.