(6er) High Performance Computing for Engineered Human Health Systems

My research uses parallel simulations to solve biological fluid and solid mechanics problems involving drug design and delivery.  High performance computing enables the representation of important physics that can only be faithfully captured by going beyond the approximations needed to make classical theories tractable.  The use of high performance computing to enable the modeling of complex behaviors involving multiple layers of physics and many bodies allows us to make better predictions in pharmaceutical design, reducing the development time and cost of new medicines.

During my Ph.D. at Stanford University as a student of Eric Shaqfeh, I used computational simulations to examine the behavior of soft, deformable vesicles in fluid flow. Understanding the physics of vesicles in flow is essential for the use of artificial vesicles as bio-compatible drug delivery agents because it allows us to achieve better targeting of delivery and to improve the yield in the production and processing of these medicines. My research incorporates subdivision surface techniques originally developed for the field of computer graphics to resolve the bending forces on the vesicle mesh with unprecedented accuracy.  The use of Loop subdivision surfaces has enabled the investigation of realistic 3D vesicles, which are far from the spherical limit. Most notably, I am able to simulate dumbbell instabilities in long tubular vesicles and match my simulations to previously published experimental literature. I have applied similar boundary integral techniques to suspensions of platelets and red blood cells and demonstrated that the volume fraction of red blood cells influences the margination of platelets and their ability to adhere to the wall of a microchannel.  These simulations demonstrate that artificial blood additives can aid the formation of blood clots through mechanisms other than the explicit mimicry of platelet functionality.

As a postdoc at UT Austin with Roger Bonnecaze, I have investigated the behavior of vesicles and cells in highly confined environments such as tight tubes and post arrays.  These simulations are used to determine the mobility of  leukocytes and cancer cells that will be incorporated into a large scale simulation of cancer cell migration, tumor growth and metastasis.  I have also used my expertise in high performance computing to accelerate simulations of the dynamics of 1000s of drops in template filling in UV roll-to-roll nanoimprint lithography.  The multi-scale model for this simulation is based on lubrication and membrane theory and includes the capillary elastohydrodynamics to study the coupled fluid-structure interaction of the flexible substrate as drops merge and fill features on the template.  By being able to simulate over a thousand drops instead of dozens, my simulations can make predictions about the maximum throughput and residual layer thickness of these nanomanufacturing processes as well as determine optimal drop patterns and quantify the effects of error tolerances in drop placement.

My future research vision uses high performance computing to build multi-scale models that help us understand the physical processes behind cancer metastasis.  Furthermore, I will use the insight gained from these models to design synthetic blood additives and to construct lab-on-a-chip medical analysis devices for detecting and manipulating cancer cells and blood borne pathogens.  Through the NSF XSEDE supercomputing program, I will continue to be eligible for time on the computer clusters I am currently using when I become faculty at another institution.  By bringing in these supercomputing resources, I will be able to establish my new computational research group swiftly and expediently.