使用分布計算及平行溫度方法模擬蛋白質摺疊; Distributed Computing And Parallel Tempering Simulation of Protein Folding

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/7327

Title:	使用分布計算及平行溫度方法模擬蛋白質摺疊;Distributed Computing And Parallel Tempering Simulation of Protein Folding
Authors:	羅嘉琳;Jia-Lin Lo
Contributors:	物理研究所
Keywords:	平行溫度法;分布計算;蛋白質摺疊;protein folding;distributed computing;parallel-tempering
Date:	2005-05-25
Issue Date:	2009-09-22 10:55:22 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	Trp-Cage 是一個由20個胺基酸組成的小蛋白質，我們使用分子動力學軟體--- GROMACS 在電腦上進行摺疊模擬，在參數設定上，我們使用真實水溶液的模型以及Gromos 43a1的力場參數，而平行溫度法也在此研究中被使用來增加尋找可能結構的速度，平行溫度法的設定中，我們使用26個溫度( 250K~478K)當作交換的依據，在GROMACS進行一段短時間的模擬之後依據公式來計算機率並決定是否交換彼此溫度，在本文的研究中，我們利用平行溫度法收集到104000個不同結構，總模擬時間為4.16 μs，除了蛋白質折疊的研究之外，另一個主題是關於分布計算環境的建構，我們取名為Protein@CBL，所有分子動力學計算皆在這個平台上進行，相關的架構及流程說明在本文中皆有敘述。此蛋白質模擬的初始結構為extended structure，針對摺疊過程中的一些參數加以討論，例如總能量，位能，機率分布，首次摺疊時間，radius of gyration，鹽橋作用，親水厭水的影響以及定壓熱容的變化，大部分的參數皆是對應於RMSD的變化，其中需要特別指出的是：當RMSD為0.7 nm、0.5 nm及0.35 nm時，其對應的首次折疊時間分別為0.6 ns、8 ns及20 ns，其顯示出這個蛋白質的折疊過程可以依據這三個RMSD值分成四個階段：extended state、molten globule、free-energy barrier 及native like，在摺疊過程的初期(extended state)，構型主要是受到牛頓力學的影響，當進入第二階段(molten globule)，力學的影響力逐漸減少而Entropy的影響逐漸增加，當RMSD ~ 0.5nm時，親水厭水的作用使得結構趨近穩定，不會再有太大變化，進入第三階段時，Entropy的作用變得相對重要，最後進入Native like，最佳結構 (RMSD = 0.23 nm) 也在經過152 ns的模擬之後出現，但是最佳結構卻不相對於最低能量，從本文的分析中可以估計，如果想得到更佳的結構 (RMSD < 0.2 nm)，總模擬時間需要約10 μs，Extended state的位能大約比平均值大540 KJ/mol，Molten globule的位能大約比平均值低 450 KJ/mol Trp-cage is a mini-protein composed of 20 residues. Here its folding is simulated in computer using the method of molecular dynamics with the aid of the software package GROMACS. Simulation was conducted in explicit solvation model and the GROMOS 96 43a1 force field was used. The parallel-tempering, or replica-exchange, algorithm was employed to enhance the efficiency of the exploration of conformational space. Folding of replicas of the peptide were simulated in twenty-six temperatures ranging from 250K to 478K are temperatures of pairs of replicas were swapped at certain time intervals according to a fixed rule. A total of 104,000 conformations were collected for analysis, representing an accumulated simulation time of 4.16 μs. A part of the present project is the construction of a distributed computing facility, called Protein@CBL, on which simulations reported here were conducted. Details on both the hardware and software aspects of Protein@CBL are described here. All simulations began with the Trp-cage replica being in an extended state. A number of parameters measuring the folding state of the replica, including total energy and potential energy, state probability, first arrival time, radius of gyration, separation of hydrophobic and hydrophilic residues, salt bridge bond length, and heat capacity were computed. In many cases these were expressed as functions of RMSD, the rms positional deviation relative to the native conformation. The three values, RMSD ~0.70, 0.50, and 0.35 nm, with first-arrival times of 0.6, 8 and 20 ns, respectively, demarcated the folding into four regimes: extended state, molten globule, free-energy barrier, and native-like. Folding was essentially kinetic in the extended state regime, dominated by kinetics but impeded by rapid loss of entropy in the molten globule regime and largely entropic thereafter. At RMSD ~0.50 nm the hydrophilic and hydrophobic residues became cleanly separated into an outer shell and an inner core. The most native-like state had a first-arrival time of 152 ns and an RMSD of 0.23 nm. The lowest energy state given by the GROMOS force-field is probably not native-like and the average time needed to fold to a state with RMSD<0.20 nm is estimated to be 10 μs. An extended state was about 540 kJ/mol above the average potential of a molten globule and a lowest energy state was about 450 kJ/mol below.
Appears in Collections:	[Graduate Institute of Physics] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...