FINDING PROTEIN SEQUENCE MOTIFS--METHODS AND APPLICATIONS

在过去的几年里，基因组序列和蛋白质结构的快速积累伴随着序列数据库搜索方法的重大进步。NCBI开发的强大的位置特定迭代BLAST(PSI-BLAST)方法形成了我们在蛋白质基序分析方面的工作基础。开发了一种新的PSI-BLAST应用模式，包括通过重复PSI-BLAST迭代来穷尽数据库搜索，以收敛于新发现的蛋白质家族成员，并在自动过程中实现。另一个新的程序，Impala，是PSI-BLAST方法的逆转，允许人们通过使用单个蛋白质序列作为查询来搜索蛋白质家族图谱的库。这些方法被应用于几类蛋白质结构域的系统分析。研究表明，在古生菌和/或细菌中可以检测到一些以前被认为是真核生物特有的信号域。通过将结构域检测和跨基因组比对相结合，将这些结构域分为祖先结构域和水平转移结构域。详细研究了组成修复系统和程序性细胞死亡系统的蛋白质结构域的进化历史。此外，对古生菌基因组中编码的DNA结合结构域进行了彻底的研究，结果表明，古生菌中这些结构域的谱系类似于细菌，但不同于真核生物。许多先前未被发现的结构域和蛋白家族被发现，包括参与多种酶的变构调节的ACT结构域多用途配体结合模型，以及与动物谷氨酰胺转氨酶同源的细菌、古菌和真核生物预测的蛋白酶超家族。-蛋白质序列基序、迭代数据库搜索、折叠识别、多比对

In the last few years, rapid accumulation of genome sequences and protein structures has been paralleled by major advances in sequence database search methods. The powerful Position-Specific Iterating BLAST (PSI-BLAST) method developed at the NCBI formed the basis of our work on protein motif analysis. A new mode of PSI-BLAST application which includes exhaustive database search by repeating PSI-BLAST iterations to convergence with newly identified protein family members was developed and implemented in an automatic procedure. Another new procedure, IMPALA, is a reversal of the PSI-BLAST method and allows one to search a library of protein family profiles by using an individual protein sequence as a query. These methods were applied to the systematic analysis of several classes of protein domains. It was shown that a number of signaling domains previously considered to be specifically eukaryotic are detectable in archaea and/or bacteria. By combining domain detection with a cross-genome comparison, these domains were classified into ancestral and horizontally transferred ones. The evolutionary histories of protein domains that comprise the repair systems and programmed cell death systems were investigated in detail. Also, the DNA-binding domains encoded in archaeal genomes have been thoroughly studied resulting in the demonstration that the repertoire of such domains in archaea resembles that in bacteria but not in eukaryotes. A number of previously undetected domains and protein families were discovered including the ACT domain  multipurpose ligand-binding model involved in allosteric regulation of avariety of enzymes and a superfamily of predicted protease from bacteria, archaea and eukaryotes that are homologous to animal transglutaminases. - Protein sequence motifs, iterative database search, fold recognition, multiple alignment