蛋白質與配體之間親和力之分析

Abstract

預測蛋白質與配體之間結合親和力對於藥物發展是一件很重要的事情，在此研究中，我們從五個構面來分析影響親和力的因素，這五個構面分別是，蛋白質與配體間的交互作用、蛋白質特性、配體的結構與物理化學性質、金屬與配體的鍵結與水的影響。並且從這五個觀點產生八十七種影響親和力的因素敘述子，再使用逐步迴歸法一項項選出與結合親和力有關的因素敘述子，建立一個新的計分函數來計算蛋白質與配體間親和力。逐步迴歸法在891筆的訓練資料中皮爾森相關係數為0.612，選出來七項因素敘述子中有六項用來解釋蛋白質與配體結合過程中各種交互作用的影響，如凡得瓦力、金屬與配體的鍵結、水的影響、與配體構型改變的懲罰，第七項則是具有氫鍵又是高保留性氨基酸的數目，我們設計了一個實驗證明在高保留性氨基酸上的氫鍵比在低保留性氨基酸的氫鍵更為重要。最後用來預測蛋白質與配體間親和力的計分函數總共包含了七個敘述子，此計分函數測試在一個獨立98筆的測試資料中的皮爾森相關係數為0.601，每一個因素敘述子的係數代表著他們對於親和力的影響，從整個結果來看，此新發展的計分函數將有助於虛擬藥物篩選與藥物設計的研究。

The prediction of the binding affinity of protein-ligand complexes is important for drug discovery. In this study, we have analyzed the descriptors, which affect the binding affinities of protein-ligand complexes, from five dimensions, including protein-ligand interactions, protein properties, structural and physicochemical descriptors of ligands, metal-ligand bonding, and water effects. Based on these dimensions, we generated 87 descriptors and used stepwise regression to select seven of these to develop a new scoring function. The correlation between predicted binding affinities and experimental values is 0.612 on a training set with 891 protein-ligand complexes selected from Protein Data Bank (PDB). The seven descriptors we selected include six general terms, i.e. van der Waals contact, metal-ligand bonding, water effects, deformation penalties upon the binding process, and one specific term, which is the number of highly conserved residues with hydrogen bonding. We had designed an experiment to prove that the hydrogen bonds in highly conserved residues are more important than ones in low conserved residues. This descriptor was termed “conserved region capture”. The final scoring function which contains seven descriptors is tested by the independent testing set of 98 protein-ligand complexes and the correlation between predicted binding affinities and experimental values is 0.601. The coefficient of each descriptor represents its influence on binding affinities of protein-ligand complexes. The experimental results show that our new scoring function for the prediction of binding affinity is useful for virtual screening and drug design.