Elucidating essential targets in pharmacologically relevant system models.

详细信息   

• 作者：Anderson ; Abraham Antonio.
• 学历：Doctor
• 年：2002
• 导师：Hunt, C. Anthony
• 毕业院校：University of California
• 专业：Engineering, Biomedical.;Computer Science.;Health Sciences, Pharmacy.
• ISBN：0493738983
• CBH：3058750
• Country：USA
• 语种：English
• FileSize：8021872
• Pages：231

文摘

Large-scale and complex biological system models can be used to simulate and mimic normal and pathologic body function. Putative pharmacological agents can be “screened” in silico for greatest effect by adjusting model parameters at postulated and known drug target sites to match experimental data. Once the simulation is set in motion, the effect can be monitored and the drug's effectiveness evaluated. This document describes how these models can be used to find potentially optimal intervention sites via quantitative and qualitative graph theoretic techniques. A major benefit of this method is its computational objectivity in analyzing large and highly connected systems. It is also much faster than a complete, classical sensitivity analysis of the model. Several systems are analyzed for modular substructure and the elements of those systems are prioritized as potential therapeutic targets. These are blood coagulation, human obesity, and bacterial metabolism. Predictions are validated with information about known therapeutic targets, essential genes and metabolites. For quantitative models, a sensitivity analysis is done to further validate prioritizations.;As we emerge from the genomic era, during which whole genomes have been completely sequenced, data mining is being used to elucidate interactions between genes and proteins. The resulting relational models are expected to be quantitatively fleshed out, and used to adequately predict biological phenomena. Biological and pharmacological system models are not new, but as a consequence of high throughput, robotic data generation technologies, they are necessarily rapidly expanding in size and getting more complex. A single researcher can no longer conceptualize even a fraction of the dynamical interactions within such models; much less judge the full effect each element may have on the system. Short of a tedious full-scale sensitivity analyses, guesses about which element or set of elements is likely to have a desired, widespread effect will be based on trial-and-error, and thus biased. The techniques employed hearken from computer network planning and parallel processor load balancing, and should return comprehensive and utterly objective results. These results may contain conceptual and computational artifacts and so were qualitatively evaluated and iteratively refined. Essentially, these techniques filter the system and present those elements that serve to hold the system together, and/or prevent chaotic behavior. Such critical elements are logical targets for evaluation as potential therapeutic intervention loci. Thus, models once used to test hypotheses, can serve double-duty by generating hypotheses. This sort of computer aided target location coupled with current methods for evaluation, are expected to give investigators the added edge of objectivity married with speed.;Scientists and engineers require new techniques and tools for exploring, visualizing, and understanding the limits and relative properties of large complex biological systems. The approach described here may be such a technique.