BIOLOGICAL COMPUTATION AND VISUALIZATION CENTER
RESEARCH


Genomics Fluid Dynamics in Living Tissues
Vision Parallel Computations
Data Mining Immersive & Interactive Vizualization

BCVC
...........................................................................................................................................
Genomics:
      Innovative analytical techniques, seminal algorithms, and massively parallel computers that can deal with the large quantity of accumulating genomic data and make full use of its potential are being developed to make rapid progress in: 
  1. Estimating phylogenetic relationships among organisms.
  2. Analyzing and modeling evolutionary processes from sequence data, including protein evolution, evolution of regulatory elements, and mobile element spread; 
  3. Predicting structural and functional features of genes; 
  4. Understanding the genetic and mechanistic bases of complex traits.
  5. Inferring population structure. 
Fluid Dynamics in Living Tissues:
     Modeling fluid flow and particle transport in biological systems is a major challenge influenced by geometrical complexity, length scales ranging from submicron to a few centimeters, deformable surfaces, and tissue components capable of interacting with fluid flow and particle movements.  Tracking fluid and particle movements in living tissues is confounded by depth of field, tissue opacity, spatial resolution, and temporal resolution of events, yet movement at the length scales involved is extremely important for many biological processes.  Examples of such processes include fluid outflow from the eye and optic neuron, fluid circulation and particle capture in gills of bivalves, and ciliary/particle interactions at variable beat frequency and particle sizes. These problems need to be addressed by integrating image capture and data processing with physical parameter-based calculations.  We are developing large-scale fluid-dynamic models, which encompass Direct Numerical Simulations, Large Eddy Simulations, and Reynolds Averaged Navier-Stokes calculations.  Vision:
     The vision group at the BCVC is developing new anatomical and morphological methods to solve problems related to fluid outflow and optic nerve axon outflow from the eye.  Optical sections of living eyes are recorded at the trabecular meshwork site and the optical nerve axon site.  In the first site, fluid outflow is monitored with a white light, real-time confocal microscope to visualize fluorescent beads moving in the aqueous flow. The data will be projected into the CAVE to allow precise manipulation of the structure-function real-time outflow mechanism of eye fluids.  The CAVE will also be used for microtomographic imaging of the lamina cribosa and its possible deformations to identify the mechanisms causing optic nerve damage under conditions of high intraocular pressure.  The vision group is also addressing the costly public health problem of vision loss from diabetes by rapidly discriminating large numbers of image files and patient data transmitted from remote clinics through the reduction of images by wavelet transforms. 
     Some of the BCVC researchers are interested in basic understanding of retinal function.  Elucidating the role of calcium buffering in amacrine cell function depends on the development of feature extraction algorithms to define the critical elevation and depletion events during sustained calcium elevations with particular emphasis on the relationship between cytosolic and mitochondrial calcium in generating interneuronic signals. Parallel Computations:
      We are designing multiresolution Grid algorithms for a variety of grand challenge biological applications.   The algorithms feature node-level (e.g., cache conscious algorithms) as well as wide-area-network-level performance optimization, load balancing, and latency tolerance.  The multidisciplinary simulation software will be wrapped by optimized parallel codes into an object-based metasystem such as Legion. Data Mining:
     Data-mining research focuses on the development of techniques to extract and combine knowledge from vast isolated datasets.  This requires standards for data integration and mechanisms for cross-referencing and checking of data.  Images and datasets from experiments and simulations are being organized as a distributed database system to avoid congestion associated with centralized approaches.  Intelligent software agents will interpret, redirect or fuse users̀ queries and results.  The system will also feature context-based image query with a simple graphical user interface, and will  have built-in intelligence with pattern recognition algorithms that will run on multiple parallel computing platforms.  Learning algorithms will be designed to classify features that are far too faint for visual classification. Immersive and Interactive Visualization:
Experimental and simulation data from biological applications will be visualized in a semi-immersive ImmersaDesk and a fully immersive CAVE.  Computations related to rendering will be performed on the Grid with the Globus metacomputing toolkit.  Visualization tools will be integrated into experimental research to enable researchers to interact with large datasets in real time even at distant locations.  Subtasks will be multithreaded, data will be compressed, and then transmitted to an ImmersaDesk or CAVE via multiple data channels, each providing information at a different level of detail and frequency to achieve real-time response.  Parallel hierarchical approaches will be developed for efficient culling, and adaptive control models will be developed to manage the trade-off between image fidelity and frame in shared virtual environments.  User friendly interfaces will be designed, so that researchers from different disciplines can use the resultant visualization framework.