藉由序列與結構比對辨認關係較遠的同源序列

Abstract

RDHSP為一蛋白質摺疊辨識法,其目的為給予一個序列然後計算出這個序列最有可能摺疊成為哪一個結構，胺基酸序列和立體結構在空間中的位置做排列，藉著適當的計分方法，計算這樣排序的得分，憑著得分的高低，判斷序列是不是會摺疊成這樣的立體結構，這種計算序列與結構之間的排序過程稱也稱作為Threading。

RDHSP利用待測蛋白質在19種「環境」(environment)( 二級結構, 量測蛋白質中被包埋起來的side chain面積疏水性及非疏水性,接觸能量,鄰近的胺基酸個數)作用的結果來分別描述此一蛋白質不同部位的結構，在此所提到的「環境」一詞，係指不同蛋白質residue間的接觸形式,接著建立在20種胺基酸在不同environment中的觀察結果，將所得資料與已知蛋白質結構或序列的資料庫比對取一胺基酸序列對上述3D profile進行排序(alignment)，比對的計分標準則依據此一序列和3D profile中所描述的結構相容性高低來判斷。

The RDHSP (Recognize Distant Homologues by Sequence-struc- ture Profile com- parison) is the protein fold recognition using the threading method. For a given target protein seq- uence and a template structure, RDHSP guarantees to find a globally optimal threading alignment between the two. RDHSP is based upon the premise that structure is better conserv- ed than sequence. Every residue in a protein tertiary stru- cture exists in a particular environment that can be descr- ibed by features such as main-chain conform- ation, solvent accessibility, and contact energy, contact residue numbers. RDHSP empl- oys environment-specific scoring table that of- fer a more precise and discriminating measure of substitut- ion probabilities. Compare with the popular PSI-BLAST, RDHSP is 4.6%, 22.2% and 21.6% more sensitive in detecting the family, superfamily, fold simila- rities in the Lindal benchmark.