A u t o m a t e d Biomodeling on the W o r l d - W i d e W e b Charles E. Harless, Joseph J. Distefano HI University of California, Los Angeles, California {charless, joed} @cs.ucla.edu Project Overview Mathematical modeling can play an important role in developing understanding about real systems. However, modeling methodology can be highly technical and mathematically complex, effectively rendering many such models unlikely candidates for the routine toolbox of most life scientists. This presents a unique problem space for biomodelers for providing software for modeling and providing proper education and support for the software techniques. One important technique used is data fitting of multi-exponential models of different dimension to small data sets, the kind typical of kinetic experiments in the life sciences. DIMSUM, an acronym for DIMension of a SUM of exponentials, is a highly automated expert system for fitting multi-exponential models of increasing dimension to time series data. It involves selection of the best-fit candidate model based oh a user-modifiable and weighted decision tree of statistical criteria for model discrimination. The model is fitted using a rough-cut curve-peeling algorithm to obtain initial parameter starting values and space boundaries. Then, the iterative gradient search algorithm is run with these as default values, with automatic adjustments of the search space boundaries when needed, rapidly achieving weighted least squares estimates. This process is repeated for all desired candidates with options for monothru 4-exponetial models. Comprehensive statistics for evaluating and comparing different order candidate models as well as a rule-based advisory subsystem provide an "experts" interpretation of the results including advice on the best-fit candidate model. Sounds easy, but it's not. Modeling programs are highly technical, DIMSUM being no exception. But DIMSUM is probably the easiest to use for this class of models, which incidentally is among the most widely used in the life and physical sciences for kinetic data analysis. This project concerns development of a new research software package for the World-Wide Web. Up to now, a researcher needs an individual copy of DIMSUM on his or her own computer as well as support to learn how to use it~ This is where the World-Wide Web comes in, and this is new territory for modeling. In the past, the web has been an undeveloped resource for biomodelers. This project aims to provide a growing resource for biomodeling on the World-Wide Web. Our developing system will be interactively available online, in real time, to all researchers with a basic Mac or PC connected to the Internet. The new software, entitled W3DIMSUM, will ',runfrom our website, resident on a server in the Biocybernetics Laboratory at UCLA (http://www.biocyb.cs.ucla.edu). The system, written in Java, C++, and Fortran, allows interactive data fitting and model discrimination over the Internet. Modelers input data and fit the data using the two algorithms built into DIMSUM. The algorithms used include the classical curve-peeling algorithm and the Marquardt-Levenberg algorithm, both of which have been thoroughly tested and proven to be valid model fitting routines. Since these algorithms are numerically intensive, we are implementing a distributed system to allow numerical processing to take place on our server. Only the user interface will be run on the client machine, downloadable through a Java Applet. This distributed system will allow future algorithms 'to be implemented as well as allowing other software to connect to the biomodeling resource. This system is a beginning to creating a large resource of implemented algorithms and software for biomodelers interested in learning and using proven modeling techniques.

/lp/association-for-computing-machinery/automated-biomodeling-on-the-world-wide-web-lO2BZ7JUFr