PHYSIOLOGICAL MODEL FOR GENE REGULATION IN DROSOPHILA

描述：这项提议旨在制定一项战略，以 许多基因之间的相互作用，这些基因决定了一种 在空间和时间上的有机体。与传统方法不同的是 分子生物学，它试图在 分子术语通过一次研究一个或最多几个基因，这 提案的重点是互动的综合网络。该方法 建议有三个主要要素。第一，计算机想象技术 将被采用和细化，以产生高质量的集中度 基因产物作为时间与空间的函数的测量 单个原子核的分辨率。这涉及到荧光测量。 带有针对每个相关基因产物的抗体和计算机 增强图像以锐化核轮廓并将其带入 一个通用的寄存器。 其次，将开发一种并行模拟退火法，以适应 对表达式数据建立数学模型。此案的进展情况 都是针对提高速度的目标，这与常规 模拟退火法的速度非常慢。要达到合适的效果 在实际应用中，用这种方法需要使用一台超级计算机。 第三，该数学模型是对传统神经网络的模拟 网络模型。对各种基因的表达变量进行加权 并将其相加，并通过非线性阈值装置来 计算出给定基因产物的合成率。还有一个 比例损失项和涉及扩散的交换项 原子核。任何基因产物的变化率由 这三个术语的组合。当这三个元素是 成功组合后，结果是一组提供最佳权重的权重 与数据相适应。这些权重然后被用来推断出 网络中各基因之间的相互作用。一直以来的模式系统 被选为论证这一战略的可行性的是 果蝇的片段基因。

DESCRIPTION: This proposal aims to develop a strategy for characterizing the interactions among the many genes that determine the properties of an organism in space and time. In contrast to the conventional approach of molecular biology, which attempts to characterize these interactions in molecular terms by the study of one or at most a few genes at a time, this proposal focuses on the integrated network of interactions. The approach proposed has three major elements. First, computer imagining techniques will be adopted and refined to generate high-quality concentration measurements of gene products as a function of time with a spatial resolution of a single nucleus. This involves fluorescence measurements with antibodies specific for each of the relevant gene products and computer enhancement of the images to sharpen the nuclear profile and bring them into a common register. Second, a parallel simulated annealing algorithm will be developed to fit a mathematical model to the expression data. The developments in this case are directed toward the goal of enhancing the speed, which with conventional simulated annealing algorithms is characteristically slow. To achieve fits in practical time with this approach requires the use of a super computer. Third, the mathematical model is an analog of the conventional neural network model. The expression variables for the various genes are weighted and summed, and the sum is passed through a nonlinear threshold device to produce the synthesis rate for a given gene product. There is also a proportional loss term and an exchange term involving diffusion between nuclei. The rate of change in any gene product is determined by the combination of these three terms. When these three elements are successfully combined, the result is a set of weights that provides the best fit to the data. These weights are then used to infer the pattern of interactions among the genes of the network. The model system that has been selected to demonstrate the feasibility of this strategy is the network of segmentation genes in Drosophila melanogaster.