a national resource for computational science education

HOME BWSIP Shodor Blue Waters

View Position
Position TitleNovel Mechanism of Blood-Clotting Could Revolutionize Stroke and Heart Attack Treatments
SummaryDespite roughly $3 billion of annual expenditure on heart-attack and brain-stroke research, these diseases that are both caused by break-ups of blood clots, remain the 1st and 4th leading cause of death in the US, respectively (killing ~750k Americans every year). Our group is combining Nobel Laureate-developed Lattice Light Sheet Microscopy (available to only ~20 other labs in the world) with parallelized image-based Lattice Boltzmann Method computational fluid dynamics in order to investigate a novel mechanism of blood clot formation. The success of this project could lead to a novel class of anti-thrombotic drugs that could selectively dissolve just the dangerous part of the blood clot, while leaving the useful one intact.
Job DescriptionThe student will:

- Develop custom Matlab (or OpenCV, if Matlab is unavailable) codes that implement HDF5 in order to parallelize 3D rendering of microscopy images by integrating the code's output with Visit and Paraview packages using the XDMF standard.

- Port the microscopy data obtained from in vivo experiments for use with our in-house Lattice Boltzmann Method (LBM) code, and run the CFD simulations in order to investigate how blood clot formation occurs in vivo. This will involve unix shell scripting in order to automate many of the job submission and post-processing tasks, as well as merging data formats between the microscopy and the LBM codes.

- Analyze and data-mine the results in order to deduce the details of the blood clotting mechanism, and identify novel drug targets for dissolving the clots more efficiently. We have terabytes of microscopy data that needs to be analyzed, so the student will write scripts in order to extract insights into haemodynamic environment experienced by the growing blood clots in order to identify what factors lead to their break-up and how to circumvent them.

- If there is time, assist with optimization of the LBM code's parallelization. Although the code is mature, there are several spots that still need to be improved. These tasks are small enough to handle by a motivated undergraduate student, and they will help her to learn Fortran, MPI and parallel I/O while at the same time improve the HPC resource use efficiency of our code. For example, a "load-balancer" was developed by XSEDE's ECSS Extended Collaborative Support Services, which minimizes the MPI communication in a particle code that simulates drug transport to the blood clot. I never had the time to incorporate it into my code, so the undergrad intern can assist me with tasks like this.

-At the end of the internship, I plan for her to submit her findings in a 1st author publication to the Blood journal.
Use of Blue WatersWe have been using a combination of XSEDE (allocations #: TG-BCS170001 and TG-BIO160074) Texas Advanced Computing Center (Stampede 1 & 2, and Ranch), U of Oklahoma Supercomputing Center (OSCER), and local NJIT supercomputer Kong. However, our XSEDE allocations will fully run out by the time of this scholarship, so I will expand the student's learning experience by exposing her to a new environment - the Blue Water's HPC resources.

We will use the Blue Water's computing resources to run the image-based Lattice Boltzmann CFD jobs, store the large images created by the microscope (that serve as input to the CFD code) and even larger results that will be generated by the CFD code on Blue Water's tape storage. Finally, we will use Blue Waters GPU nodes in order to visualize the results using either Paraview or VisIt open source software. The student will also use the Blue Waters resources in order to optimize the parallelization of the Fortran/MPI Lattice Boltzmann CFD and Matlab image processing codes.

Currently, NJIT does not have any supercomputing courses. Hence, I as a tenure track professor, I am trying to bring HPC education to our campus. I believe that the use of the Blue Water's resources that I described above will be synergistic with my efforts here at NJIT.
Conditions/QualificationsMust be an undergraduate student at NJIT with experience in:

- Lattice Light Sheet Microscopy image processing
- Lattice Boltzmann Method parallel computing using Fortran and MPI
- HDF5 and Matlab
Start Date05/31/2018
End Date05/31/2019
LocationOtto H. York Department of Chemical Engineering
New Jersey Institute of Technology (NJIT)
Tiernan Hall, Suite 150
University Heights
Newark, New Jersey 07102
Migle Surblyte