Translog: How estimate the function of a gene whose seuqnece is unique

本研究提出了一种利用网络结构相似性来估计基因功能的方法，而不是目前主要使用的序列相似性来估计基因功能。为此，我们开发了一种比较全局网络结构的方法ANGIE。给定两个不同的网络，ANGIE试图在每个网络上找到一个简短的网络结构，使两个简短的图彼此相似。由于我们最初的实验结果表明，比对结果与生物学知识相差甚远，因此我们对ANGIE进行了改进，从而给出了两个物种之间基因的先验相似性知识，如序列相似性。我们将ANGIE应用于酵母和果蝇网络，这些网络是由蛋白质-蛋白质相互作用或表达相似性产生的，结果表明基因具有新的功能。

This study developed a method to estimate the functions of genes using similarity of network structure, not using sequence similarity which has been currently mainly used. To this purpose, we developed the method ANGIE which can compare global network structure. Given two different networks, the ANGIE tries to find a brief network structure on each network so that the two brief graphs are similar to each other. Since our initial experimental result showed that the aligned result is far from biological knowledge, we improve the ANGIE so as to give prior similarity knowledge of genes between two species, such as sequence similarity. We applied the ANGIE to yeast and drosophila networks, generated from protein-protein interactions or expression similarity, and the result suggested new functions of genes.