Journal of Computational Chemistry A Toolkit to Assist ONIOM Calculations For Peer Review Journal: Journal of Computational Chemistry Manuscript ID: Draft Wiley - Manuscript type: Software News and Updates Date Submitted by the Author: Complete List of Authors: Tao, Peng; Wayne State University, Department of Chemistry Schlegel, H. Bernhard; Wayne State University, Department of Chemistry Key Words: QM/MM, ONIOM, charge fitting, efficiency John Wiley & Sons, Inc. Page 1 of 20 Journal of Computational Chemistry 1 2 3 4 A Toolkit to Assist ONIOM Calculations 5 6 7 8 9 10 11 Peng Tao, and H. Bernhard Schlegel * 12 13 14 Department of Chemistry, Wayne State University, 5101 Cass Ave, Detroit, Michigan 48202 15 16 17 [email protected] 18 For Peer Review 19 RECEIVED DATE 20 21 22 Correspondence to: H. Bernhard Schlegel, Department of Chemistry, Wayne State University, Detroit, 23 24 Michigan 48202. Email: [email protected] 25 26 27 This article contains supplementary material available in the online version of this article. 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1 John Wiley & Sons, Inc. Journal of Computational Chemistry Page 2 of 20 Abstract: A general procedure for QM/MM studies on biochemical systems is outlined, and a 1 2 3 collection of PERL scripts to facilitate ONIOM type QM/MM calculations is described. This toolkit is 4 5 designed to assist in the different stages of an ONIOM QM/MM study of biomolecules, including input 6 7 file preparation and checking, job monitoring, production calculations, and results analysis. An iterative 8 9 10 procedure for refitting the partial charges of QM region atoms is described and yields a more accurate 11 12 treatment of the electrostatic interaction between QM and MM regions during QM/MM calculations. 13 14 15 The toolkit fully supports this partial charge refitting procedure. By using this toolkit for file 16 17 conversions, structure manipulation, input sanity checks, parameter lookup, charge refitting, tracking 18 For Peer Review 19 optimizations and analyzing results, QM/MM studies of large size biochemical systems can be much 20 21 22 more convenient and practical. 23 24 25 26 Keywords: QM/MM, ONIOM, charge fitting, efficiency 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 2 John Wiley & Sons, Inc. Page 3 of 20 Journal of Computational Chemistry 1 Introduction 2 3 Computational chemistry and biology are becoming routine methods to study many biological 4 5 processes. In computational studies of biochemical reactions involving macromolecules, one of the most 6 7 powerful tools available is combined quantum mechanics and molecular mechanics (QM/MM) 8 1-5 9 methods. In QM/MM methods, the large system of interest is divided into two (or more) parts. The 10 larger part, which is not directly involved with the chemical reaction, is treated at a low cost level of 11 12 theory, such as molecular mechanics (MM). The smaller part where the chemical reaction occurs is 13 14 treated with a more accurate level of theory, such as quantum mechanics (QM), with a higher 15 16 computational cost. Using this scheme, the entire system, e.g. an enzyme with substrate and solvent 17 molecules, can be treated computationally. The reaction of interest in the active site of the enzyme can 18 For Peer Review 19 be studied at an appropriate level of theory with an affordable computational cost. 20 21 Although different cases may vary, the general procedure for a typical QM/MM study of a 22 23 biochemical system can be described as follows. First, reliable experimental structures, usually crystal 24 25 structures, are identified and preprocessed. The preprocessing includes adding hydrogen atoms, 26 determining the most probable protonation states of titratable amino acids, add counter ions and solvent 27 28 molecules, geometry optimization with MM and equilibration by molecular dynamics (MD) using MM. 29 30 Then, representative geometries are selected from the MD simulations and are subject to optimization 31 32 again before the QM/MM study is initiated. Third, QM/MM geometry optimizations are set up from 33 selected MD geometries and conducted for the initial structures, usually reactant complexes in the 34 35 reaction sequence under study. Fourth, transition states (TS), intermediate structures and products of 36 37 interest are generated from reactant structures and optimized. Constructing and optimizing 38 39 representative structures along the reaction path is the most time consuming part of the study and needs 40 41 to be carried out very carefully. Fifth, the structures identified in along reaction path are subject to 42 production calculations at a higher level of theory than used for the geometry optimizations. This 43 44 general process is illustrated in Figure 1. 45 46 Based on the mechanism of coupling between the QM and MM region, two types of QM/MM 47 6 48 methods are generally available, additive and subtractive. ONIOM is one popular and robust 49 subtractive QM/MM method. 7,8 The total ONIOM energy for a two layer system is 8 50 51 E ONIOM = E real , MM + E model , QM − E model ,MM 52 53 In a two-layer ONIOM QM/MM calculation, the real system contains all the atoms (including both QM 54 real,MM 55 and MM region) and is calculated at the MM level ( E ). The model system is the QM region treated 56 model,QM 57 at the QM level ( E ). To obtain the total ONIOM energy, the model system also needs to be 58 treated at MM level (Emodel,MM) and be subtracted from real system MM energy. 59 60 3 John Wiley & Sons, Inc. Journal of Computational Chemistry Page 4 of 20 Currently, the QM/MM methods implemented in the Gaussian package 9 are two- and three-layer 1 2 ONIOM models for calculating the energy, gradient and Hessian for geometry optimization and 3 4 vibrational frequency analysis. ONIOM calculations using Gaussian have been applied in numerous 5 10-18 6 studies in computational biology, chemistry and material science. In a subtractive QM/MM method 7 such as ONIOM, the MM force field is needed for entire system, including QM region. The MM force 8 9 field includes parameters for bond and non-bond interactions and partial charges. Due to its subtractive 10 11 nature, most bond and short-range non-bonded terms in the QM region cancel in the ONIOM QM/MM 12 13 method. 14 15 Two schemes are generally applied to treat the electrostatic interaction between the QM and MM 16 regions. 19 In the mechanical embedding (ME) scheme, this interaction is treated at MM level. In the 17 18 electrostatic embedding (EE)For scheme, Peer the partial charges Review of MM atoms are included in the Hamiltonian 19 20 of the QM region. In the ME scheme, the electrostatic interaction between the QM and MM regions do 21 22 not cancel in the QM/MM energy. The partial charges of QM atoms play an important role in this 23 scheme, therefore they need to be treated carefully in QM/MM calculations. 24 25 Since the size of biochemical systems studied by QM/MM can be rather large, some of the structure 26 27 manipulation and processing of these systems can be cumbersome. During a QM/MM calculation, it is a 28 29 good practice to monitor running jobs in terms of both energies and geometries. If problems in running 30 31 jobs can be spotted at an early stage, much time and resources can be saved. However, since the sizes of 32 structures and files in QM/MM studies can be rather large, it is not always convenient to check running 33 34 QM/MM jobs on the fly, especially geometries. Although there are many graphical interfaces available 35 36 to facilitate QM/MM calculations, the efficiency and convenience of QM/MM studies can be 37 38 substantially increased when appropriate auxiliary tools are available for job preparation, monitoring, 39 and analysis. In this paper, we describe a Toolkit to Assist ONIOM calculations (TAO). The toolkit is 40 41 written in PERL using an object oriented design and was developed during our previous QM/MM 42 18 43 studies. With these tools, QM/MM studies using the ONIOM method in Gaussian become much easier 44 45 and more efficient, especially for large enzymatic systems. To illustrate the use of the toolkit and to 46 guide a user through an ONIOM QM/MM study of an enzymatic system, we also provide a tutorial in 47 48 the Supplementary Information. 49 50 In this paper, the general procedure of a QM/MM study is discussed. The tools available for the 51 52 various stages of this study are listed in Table I. 53 54 55 56 57 58 59 60 4 John Wiley & Sons, Inc. Page 5 of 20 Journal of Computational Chemistry ONIOM input preparation 1 2 Since many large scale QM/MM studies are focused on biological systems, such as proteins, this 3 4 paper is directed mainly toward calculations of these systems. However, this toolkit can also be applied 5 6 to any other system suitable for QM/MM studies. The Protein Data Bank (PDB) is a common resource 7 for initial structures of biomolecules. The PDB file format is one of the most commonly used formats 8 9 for biomolecule structure storage.