Finding Optimal Models for Small Gene Networks
Finding Optimal Models for Small Gene Networks
复制标题
DOI：
10.1142/9789812704856_0052
复制
发表时间：
2003-12
影响因子：
--
通讯作者：
Sascha Ott;S. Imoto;Satoru Miyano
中科院分区：
文献类型：
--
作者：
Sascha Ott;S. Imoto;Satoru Miyano
从微阵列数据中发现基因网络是近年来的研究热点之一。给定超指数大小的搜索空间，研究人员一直在应用启发式方法，如贪婪算法或模拟退火来推断这样的网络。然而，免疫学的准确性是不确定的，这与微阵列的高测量噪声相结合，使得很难从免疫学估计的网络中得出结论。我们提出了一种方法，找到最佳的贝叶斯网络的相当大的规模，并显示第一个结果的应用酵母数据。在消除了启发式方法的不确定性之后，可以评估不同统计模型的能力，以找到生物学上准确的网络。
Finding gene networks from microarray data has been one focus of research in recent years. Given search spaces of super-exponential size, researchers have been applying heuristic approaches like greedy algorithms or simulated annealing to infer such networks. However, the accuracy of heuristics is uncertain, which--in combination with the high measurement noise of microarrays--makes it very difficult to draw conclusions from networks estimated by heuristics. We present a method that finds optimal Bayesian networks of considerable size and show first results of the application to yeast data. Having removed the uncertainty due to the heuristic methods, it becomes possible to evaluate the power of different statistical models to find biologically accurate networks.