Computational Core

点击翻译按钮获取中文摘要

Brandeis University has a long history in computational neuroscience. This work has been characterized by the close and intimate relationships between theoretical and computational work, and many of the experimentalists on this grant (Marder, Lisman, Turrigiano, Nelson) have published numerous computational and modeling papers that have illuminated and/or driven many of their experimental findings. At present, we have a strong computational journal club that meets every Monday with attendance of about 20 faculty, postdocs, and graduate students, many of whom work in NINDS-funded labs. For example, Marder now has 4 postdocs and 3 graduate students who are doing exclusively or partially computational work, one of whom is now building models of Birren's neurons Lisman has longstanding computational projects and Paul Miller, a theorist and new hire, is already in collaborative work with Lisman, Fiser, Katz and Turrigiano. We believe that our ability to build and analyze computational models is becoming increasingly important in generating new hypotheses relevant to experimental work. We are now in an era in which reductionist approaches to biology are being supplemented by efforts to account for system behavior in terms of their underlying components. Whether we call this "systems biology" or another name, it remains the case that if one wants to understand the interaction of multiple, non-linear processes, it becomes useful, even necessary, to build models and determine which properties of the system depend on which component interactions. For example, the Birren/Marder computational collaboration arose in response of data generated in the Birren laboratory that were "begging" for a model to help understand these data and make further predictions from them. Therefore, we envision that more and more of our experimental colleagues will find it useful to have both computational expertise on campus, as well as a Core facility that will be available to run simulations, large and small. The addition of a Computational Core will make tangible our firm belief that computational and modeling work will become increasingly important to all of our experimental laboratories, and that these facilities be easily accessible and supported. Initially our computational work was done with lab-based computers, and/or small individual clusters of processors. This has two disadvantages: the relatively small number of processors available to any given investigator, and too much time being spent in maintaining too many small clusters. Consequently, a number of faculty around campus decided, with the advice and help of Dr. Steven Karel, our Biology Computational Expert, and LTS, the university's technology department, to consolidate the various small clusters found in neuroscience, biochemistry, and physics, into one large cluster. We are in the process of doing so, by taking all of our existing machines and then adding to them new hardware; we expect this process to be complete by the end of 2007. This new computational cluster will need oversight and maintenance for most effective management, and it will also need periodic replacements and upgrades. Because computational work is so integral to the experimental components of so many of us, we feel that this cluster will provide a significant asset to the NINDS-funded research we are now doing, and want to add this newly centralized facility to this Core grant. This is expecially important to maintaining consistency in the cluster maintenance and operation.