Personal Page of Eric Haseltine

Welcome to the personal page of Eric Haseltine. I am a chemical engineer by training, with a B.S. from Clemson University, a Ph.D. from the University of Wisconsin-Madison, and postdoctoral training at the California Institute of Technology. I am currently working at Vertex Pharmaceuticals in Cambridge, MA as a research scientist, where I am developing mathematical models to aid in the understanding and treatment of human disease. My research interests lie broadly in anything systems (systems biology, systems theory, systems engineering). Recent endeavors include:

1. Design principles of cell-cell communication. Autocrine signaling, the process by which cells of the same type communicate with one another, is commonly thought to coordinate population decision making. We are using synthetic gene circuits to understand how the structure of the signaling regulatory network affects this coordination. Our experimental results suggest that different architectures can either minimize or amplify the effects of cell-to-cell variations.

In collaboration with Frances H. Arnold (Caltech).

2. Systems modeling of viral infections. How do viruses propagate, and how can we control this propagation? Modeling offers one way of answering these questions. Most mathematical models of viral infections focus on either how the virus uses the resources of its host cell to make more virus (the intracellular level), or how populations of virus and host cells interact (the extracellular level). We have developed multi-scale models that incorporate both levels of information. By systematically understanding how viral infections progress, we hope to identify either new drug targets or new treatment strategies using existing drugs.

In collaboration with Jim Rawlings and John Yin (UW-Madison).

3. Approximations for stochastic chemical kinetics. In living cells or on a catalyst particle, the number of reactive molecules are typically small and countable. Chemical reactions in these systems can be modeled jumps from one discrete state to another by using discrete probabilistic models (Markov processes). These problems are typically solved using stochastic simulation due to their large size. Even these simulations can prove to be computationally expensive, so we have developed approximations using decomposition strategies and singular perturbation analysis to overcome these difficulties. We are also interested in how to perform systems-level tasks (parameter estimation, optimal control, etc.) from simulation.

In collaboration with Jim Rawlings (UW-Madison).

4. State estimation strategies for nonlinear model predictive control. To control a process given a nonlinear model, one must first identify the current state of the process from process measurements (state estimation) before deciding how to drive that process to the desired operating point. Optimization can be used to reconcile process measurements with model predictions. We have investigated one such strategy, moving-horizon estimation (MHE), for this purpose. The benefit of this strategy is that process knowledge in the form of constraints can be used to improve the estimation.

Doctoral work with Jim Rawlings (UW-Madison).

1. Design principles of cell-cell communication. Autocrine signaling, the process by which cells of the same type communicate with one another, is commonly thought to coordinate population decision making. We are using synthetic gene circuits to understand how the structure of the signaling regulatory network affects this coordination. Our experimental results suggest that different architectures can either minimize or amplify the effects of cell-to-cell variations.
In collaboration with Frances H. Arnold (Caltech).

2. Systems modeling of viral infections. How do viruses propagate, and how can we control this propagation? Modeling offers one way of answering these questions. Most mathematical models of viral infections focus on either how the virus uses the resources of its host cell to make more virus (the intracellular level), or how populations of virus and host cells interact (the extracellular level). We have developed multi-scale models that incorporate both levels of information. By systematically understanding how viral infections progress, we hope to identify either new drug targets or new treatment strategies using existing drugs.
In collaboration with Jim Rawlings and John Yin (UW-Madison).

3. Approximations for stochastic chemical kinetics. In living cells or on a catalyst particle, the number of reactive molecules are typically small and countable. Chemical reactions in these systems can be modeled jumps from one discrete state to another by using discrete probabilistic models (Markov processes). These problems are typically solved using stochastic simulation due to their large size. Even these simulations can prove to be computationally expensive, so we have developed approximations using decomposition strategies and singular perturbation analysis to overcome these difficulties. We are also interested in how to perform systems-level tasks (parameter estimation, optimal control, etc.) from simulation.
In collaboration with Jim Rawlings (UW-Madison).

4. State estimation strategies for nonlinear model predictive control. To control a process given a nonlinear model, one must first identify the current state of the process from process measurements (state estimation) before deciding how to drive that process to the desired operating point. Optimization can be used to reconcile process measurements with model predictions. We have investigated one such strategy, moving-horizon estimation (MHE), for this purpose. The benefit of this strategy is that process knowledge in the form of constraints can be used to improve the estimation.
Doctoral work with Jim Rawlings (UW-Madison).