At the Computational Neuroscience Laboratory (CNL) of Dr. Terry Sejnowski at the Salk Institute for Biological Studies in San Diego and at the University of Tennessee Computer Science Department, participants in the National Partnership for Advanced Computational Infrastructure (NPACI) have united their forces to combine two innovative computer applications, in order to conduct extremely powerful simulations of microscopic cellular processes. With the assistance of the MCell and NetSolve systems, it might even be possible to simulate the functions of an entire organism, as stated by neurobiologists in the recent issue of the enVision magazine. If simulation techniques are integrated with distributed processing workload methods, scientists are able to study in detail what exactly is happening at the level of the various brain synapses.
The MCell simulation software programme has been developed over the past six years at both CNL and the Division of Biological Sciences at Cornell University. Dr. Thomas Bartol from CNL and Dr. Joel Stiles from the Cornell Miriam Salpeter Lab have designed the Monte Carlo simulator for cellular microphysiology with the theoretical aid of Dr. Edwin Salpeter. The scientists made use of the Monte Carlo diffusion method as well as chemical reaction algorithms in 3D to allow simulation of complicated biochemical interactions of molecules inside and outside the living cells. The major goal of the researchers is to observe how neurotransmitters diffuse and activate receptors in the brain and other synapses between different cells.
Currently, the MCell application is being submitted to several specialized beta-tests. Grants to fund this project were offered by the National Science Foundation (NSF) and the National Institutes of Health (NIH). Adequate simulation however only constitutes one aspect of the problem. In order to detect interactions between millions of molecules over milliseconds or just seconds of time, requires an enormous computational processing power. It is impossible for one single computer to handle a workload that large in an acceptable amount of time. Fortunately, the issue has been solved with the implementation of NetSolve, an Internet resource management tool, which has been developed in 1997 by the NPACI partners Jack Dongarra and Henri Casanova at the University of Tennessee.
Researchers at the Oak Ridge National Laboratory were equally involved in the creation of the NetSolve client-server application, which allows users to access both hardware and software computational resources, distributed across the network. Since the programme operates as an agent, which can locate a selection of fast computers to balance the simulation workload, NetSolve offers a tremendous opportunity for researchers to remotely solve the most complex scientific problems. MCell, for instance, presents a huge number of parameters for examination, which are simultaneously being distributed by NetSolve among a wide range of computers in the best and most efficient way. If one of the runs fails, the programme automatically resubmits the task for a new trial.
The NetSolve system allows the integration of a variety of interfaces, such as Fortran, C, Matlab, Java and Perl, to enhance ease of use for the scientist. Just like the MCell software, NetSolve is funded by a grant from the NSF. The combined MCell/NetSolve solution opens up a whole new perspective of large-scale simulations, in which the researcher will spot a profusion of organism activity during a far longer period of time than ever before. Also other parts of the human anatomy than the brain synapses can be explored through high performance simulation. In addition, NetSolve can be applied to whatever research domain, which implies the use of vast networked computational power.