How accurately can we discriminate G-protein-coupled receptors as 7-tms TM protein sequences from other sequences?
How accurately can we discriminate G-protein-coupled receptors as 7-tms TM protein sequences from other sequences?
复制标题
DOI：
10.1016/j.bbrc.2005.10.123
复制
发表时间：
2005-12-23
影响因子：
3.1
通讯作者：
Shimizu, T
中科院分区：
文献类型：
--
作者：
Inoue, Y;Yamazaki, Y;Shimizu, T
在SWISS-PROT 46.0中注册的2502个具有7个跨膜（TM）片段（7-tms）的跨膜（TM）蛋白序列组包含2200个G蛋白偶联受体（GPCR），表明仅通过正确预测TM片段的数目为7-tms就可以以87.9%的可靠性检测真核基因组中的GPCR候选物。目前提出的TM拓扑预测方法的预测精度都不高，即使是最好的方法，HMMTOP 2.0，也只能达到77.6%的7-TMS序列的捕获率。有必要尽可能提高这一性能，即使只有几个百分点，以便在越来越多的新测序基因组中识别出尽可能多的新型GPCR候选基因。在这项研究中，我们提出了一个简单但有用的预测方法检测尽可能多的7-TMS TM蛋白序列的GPCR候选人在真核基因组中。这是通过采用两步预测过程来实现的。第一步涉及通过最佳预测方法（HMMTOP 2.0）收集7-tms序列，第二步涉及通过次佳方法（TMHMM 2.0）拾取剩余的7-tms序列。通过该方法，SWISS-PROT中7-tms TM蛋白序列的捕获率可以从77.6%显著提高到84.5%，并且在人类基因组（Build 35）中预测为7-tms的GPCR候选序列的数量从790（通过HMMTOP 2.0）增加到903。这790和903个候选序列分别包括SWISS-PROT 46.0中登记的717个已知人GPCR中的587个和636个，表明所提出的组合方法在检测真核基因组中的GPCR候选基因中是有效的。(c)2005年爱思唯尔公司All rights reserved.
The group of 2502 transmembrane (TM) protein sequences with seven TM segments (7-tms) registered in SWISS-PROT 46.0 contains 2200 G-protein -coupled receptors (GPCRs), indicating that GPCR candidates can be detected with a reliability of 87.9% in the eukaryotic genomes merely by correctly predicting the number of TM segments as 7-tms. The predictive accuracies of TM topology-prediction methods proposed so far are not as high as expected; even the best method, HMMTOP 2.0, can only achieve a capture rate of 7-tms sequences of 77.6%. It is necessary to improve this performance as much as possible, even if by only a few percentage points, in order to identify as many novel GPCR candidate genes as possible among the increasing number of newly sequenced genomes. In this study, we propose a simple but useful prediction method for detecting as many 7-tms TM protein sequences as GPCR candidates in eukaryotic genomes as possible. This is achieved by employing a two-step prediction procedure. The first step involves collecting 7-tms sequences by the best prediction method (HMMTOP 2.0), and the second involves picking up the remaining 7-tms sequences by the second-best method (TMHMM 2.0). By this procedure, the capture rate of 7-tms TM protein sequences in SWISS-PROT can be improved considerably from 77.6% to 84.5%, and the number of GPCR candidate sequences predicted as 7-tms in the human genome (Build 35) is increased from 790 (by HMMTOP 2.0) to 903. These 790 and 903 candidate sequences include, respectively, 587 and 636 of the known human GPCRs of the 717 registered in SWISS-PROT 46.0, demonstrating that the proposed combinatorial method is effective in detecting GPCR candidate genes in eukaryotic genomes. (c) 2005 Elsevier Inc. All rights reserved.