Enhanced Sampling And Markov State Model Analysis For Biomolecule

Biomolecules play important roles in life science.When they perform their regulatory functions,they are accompanied by complex conformational changes.However,these changes are difficult to capture by conventional dynamics method.It is necessary to use enhanced sampling methods to obtain sufficient sampling data for the analysis of the specific regulation mechanisms.Generally,there are two ways to improve the sampling efficiency.One is increasing the environment temperature of the system,and the other is appling a strong bias potential to the system.Based on previous work in the past,this paper proposes an effective sampling algorithm,called Mixing REMD.First,it imposes an adaptive biasing potential to some replicas of the molecule.These replicas sample quickly in the collective variable space.Secondly,it places more exchangeable replicas at different temperatures.These replicas generate the unbiased sampling data in the canonical ensemble.In order to improve the sampling efficiency,the states of the biased replicas are transferred to the unbiased replicas after equilibrium.Compared to other methods,Mixing REMD is more convenient to users.There is no need to prepare a static bias potential before the formal simulation.The number of the unbiased replicas are limited because they do not have to overcome the free energy barrier by themselves.After the simulation,the unbiased sampled data can be directly used for the free energy calculation on any collective variables without the additional reweighting operations.In order to implement the Mixing REMD method,we develope a fast sampling and analysis tool(FSATOOL).FSATOOL has the following features: First,the entire sampling and analysis process only requires two steps and two input files,which are very simple and convenient.Second,it is embedded in the AMBER software.Simulation with FSATOOL still contains the rich functions of AMBER.Third,it provides many analysis algorithms,including clustering,dimensionality reduction,etc.The most important point is the ability to construct the Markov State Mode(MSM)from the sampling data and extract the dominant transition pathways between the states.As a test,we apply FSATOOL in the simulation of the alanine dipeptide,the N-terminal domain of ribosomal protein L9,and the RNA hairpin.Moreover,we study the structural plasticity of an RNA motif,k-turn.K-turn can adjust its conformation between two stable states,N1 and N3,according to the environment.After an in-depth analysis,we have the following conclusions: First,when k-turn exists as an independent molecule in solvent,the stabilities of its N1 and N3 structure are close to each other.The entire molecule acts as a hinge.However,when k-turn binds to different proteins,such as protein L7 Ae and L24 e,it is trapped into N3 and N1,respectively.This finding is consistent with the experimental results.Due to the RNA-protein interaction,the free enegy barrier on the path between N1 and N3 increases than that of a single k-turn,which slows down the related reaction rate between the two states.Finally,we use FSATOOL to study the structural movement of calmodulin.Calmodulin is involved in many signal transductions in cells.However,due to the complexity of the structure,many previous works only focus on the dynamic behavior of its N-terminal or Cterminal domain in the solvent.In this work,we obtain the complete transition path of the entire protein from the closed state to the open state by the Mixing REMD method.We find that calcium ions are capable of changing the probability of distribution and transition process.It reveals that calcium ions do have an important regulatory effect on calmodulin.