搜尋相似的RNA三級子結構

Abstract

近年來人們對非編碼RNA (ncRNAs)的興趣正快速地成長，因為他們在細胞內扮演著許多重要的角色，儘管這些ncRNAs不會被轉譯成蛋白質。事實上，大多數已有的ncRNAs的功能仍是未知。如同蛋白質，一個較為可靠去決定出ncRNA功能的方法便是去分析他們的三級結構，因為分子的結構通常會比他們的一級序列在演化上還來得保守。在這方面，最近一連串的努力與研究已使得存放在PDB資料庫裡頭的RNA三級結構在數量與大小上都大大地增加。因此，發展出一個能夠快速且正確地搜尋出PDB資料庫裡結構相似的RNAs的自動化工具就顯得愈來愈重要了。在這個研究中，我們利用結構字元的方法發展出一個名叫R3D-BLAST工具讓生物學家去搜尋PDB資料庫裡與某一個RNA三級結構相似的RNAs。我們設計出R3D-BLAST背後的基本想法如下：首先，我們利用RNA核苷酸骨幹上的二個假的扭轉角(Pseudo-torsion Angles)以及親和性互動式(Affinity Propagation)的分群方法得到一個含有23個字母的結構字元集，然後再根據這個結構字元集把目前存放在PDB資料庫裡頭所有RNA三級結構編碼成一級的序列。接著我們再利用BLAST這個程式去搜尋出與qury RNA三級結構局部相似的RNAs。我們實驗的結果最後證明：我們的R3D-BLAST在識別出與qury RNA三級結構有局部相似的RNAs這方面的表現確實比BLAST還要好，而且在找出與qury RNA三級結構有整體相似的RNAs這方面的表現也比FASTR3D還好。因此，我們相信R3D-BLAST在結構生物學的研究上可以充當一個有用的生物資訊工具。

In recent years, there is a fast growing interest in non-coding RNAs (ncRNAs) because they play a lot of essential roles in many cellular processes, even though the transcripts of these ncRNAs are not translated into proteins. Actually, the function of most available ncRNAs is still unknown. Likewise to proteins, a more reliable way for determining the functions of ncRNAs is to analyze their three-dimensional (3D, tertiary) structures, because structures of molecules are typically more evolutionarily conserved than their primary sequences. In this regard, a series of recent efforts and studies has led to a substantial increase in both the number and the size of solved RNA tertiary structures deposited in the PDB database. Therefore, it has become more and more crucial to develop automatic tools that are able to fast and accurately search the PDB database for structurally similar RNAs. In this study, we have used a structural-alphabet approach to develop a web server, called R3D-BLAST, that allows biologists to search the PDB database for structural similarities of an RNA 3D structure. The basic idea behind our R3D-BLAST is as follows. We first encode all the RNA 3D structures deposited in the PDB database as 1D sequences using the structural alphabet of 23 letters, which was obtained by using the two pseudo-torsion angles of RNA nucleotide backbones and the affinity propagation clustering approach. We then apply BLAST to searching for RNA molecules whose 3D structures are locally similar to that of the query RNA. Our experimental results have finally shown that our R3D-BLAST indeed has better performance than BLAST, a famous bioinformatics tool to find homologous proteins/RNAs only based on their sequence similarity, for identifying those RNA molecules whose tertiary substructures are locally similar to that of the query RNA, as well as FASTR3D for finding those RNAs whose structures are entirely similar to that of the query RNA. Therefore, we believe that our R3D-BLAST can serve as a useful bioinformatics tool in the study of structural biology.