An application of fragment interaction analysis based on local MP2

We have developed a method named ‘fragment interaction analysis based on local MP2’ (abbreviated as FILM). This method enables us to decompose the interaction energy associated with dispersion interactions into contributions of localized occupied orbitals. In this study, the basis set dependence of the results derived from FILM was examined. The results suggested that the individual ratio of pair correlation energies of selected orbital pairs to the total dispersion interaction was almost independent of the basis set size. As an illustrative example, detailed analysis was performed on the human immunodeficiency virus type 1 protease complexed with lopinavir molecule.     

Analysis of non-covalent interactions in (bio)organic molecules using orbital-partitioned localized MP2

Localized molecular orbitals (LMO) are used as basis for an MP2 treatment (LMP2) of electron correlation energies. The major aim is an improved understanding of the non-covalent interactions in large molecules with an emphasis on intra-molecular dispersion effects. A partitioning of the inter-fragment electron correlation energy into electron pairs of different orbital type (i.e., sigma, pi, lone-pairs) is presented. The benzene dimer, 1,4-diphenylbutane conformations, and the tyrosine-glycine dipeptide are used as model systems. For the benzene dimer, comparisons with CCSD(T) data are made in order to analyse the MP2 problems for pi-pi stacking. A comparison of phenyl-phenyl interactions in the benzene dimer and for 1,4-diphenylbutane conformations reveals a very good transferability of dispersion-type contributions to binding from an inter-molecular to an intra-molecular situation. In both systems, the relative (percentage) contributions of sigma-sigma, sigma-pi, and pi-pi pairs to the total inter-fragment correlation energy is a clear signature for the binding mode (pi-stacked vs. T-shaped). For various benzene dimer conformations, we find a linear relation between the MP2 interaction energy error and the correlation contribution from pi-pi pairs. In the dipeptide, also dispersion-type electron correlations between the glycyl amino acid residue and the phenol group are most relevant for folding. This convincingly explains problems of DFT with such systems reported previously. Although in this case only one aromatic ring (and a glycyl moiety) is involved, the same sigma-sigma, sigma-pi, and pi-pi correlations seem to dominate the shape of the potential energy surface.     

Application of resolution of identity approximation of second-order M?ller�CPlesset perturbation theory to three-body fragment molecular orbital method

The resolution of identity (RI) approximation of second-order M?ller?CPlesset perturbation (MP2) theory, termed as RI-MP2, is applied to three-body fragment molecular orbital (FMO3) method. New implementation of FMO3 RI-MP2 is developed based on an efficient parallel RI-MP2 code developed recently in our group. Using this new implementation, the accuracy and computational time of FMO3 RI-MP2 calculations are assessed for water clusters, polyalanines, and proteins. The errors arising from RI-MP2 are sufficiently small in the FMO3 MP2 calculations that they give quantitative accuracy for practical chemical applications. Considerable time savings are attained in the FMO3 MP2 calculations with the application of RI-MP2.     

Comparison of ab initio and DFT electronic structure methods for peptides containing an aromatic ring: effect of dispersion and BSSE

We establish that routine B3LYP and MP2 methods give qualitatively wrong conformations for flexible organic systems containing pi systems and that recently developed methods can overcome the known inadequacies of these methods. This is illustrated for a molecule (a conformer of the Tyr-Gly dipeptide) for which B3LYP/6-31+G(d) and MP2/6-31+G(d) geometry optimizations yield strikingly different structures [Mol. Phys. 2006, 104, 559-570]: MP2 predicts a folded "closed-book" conformer with the glycine residue located above the tyrosine ring, whereas B3LYP predicts a more open conformation. By employing different levels of theory, including the local electron correlation methods LMP2 (local MP2) and LCCSD(T0) (local coupled cluster with single, double, and noniterative local triple excitations) and large basis sets (aug-cc-pVnZ, n=D, T, Q), it is shown that the folded MP2 minimum is an artifact caused by large intramolecular BSSE (basis set superposition error) effects in the MP2/6-31+G(d) calculations. The B3LYP functional gives the correct minimum, but the potential energy apparently rises too steeply when the glycine and tyrosine residues approach each other, presumably due to missing dispersion effects in the B3LYP calculations. The PWB6K and M05-2X functionals, designed to give good results for weak interactions, remedy this to some extent. The reduced BSSE in the LMP2 calculations leads to faster convergence with increasing basis set quality, and accurate results can be obtained with smaller basis sets as compared to canonical MP2. We propose LMP2 as a suitable method to study interactions with pi-electron clouds.     

Second order Møller-Plesset perturbation theory based upon the fragment molecular orbital method

The fragment molecular orbital (FMO) method was combined with the second order M?ller-Plesset (MP2) perturbation theory. The accuracy of the method using the 6-31G(*) basis set was tested on (H(2)O)(n), n=16,32,64; alpha-helices and beta-strands of alanine n-mers, n=10,20,40; as well as on (H(2)O)(n), n=16,32,64 using the 6-31 + + G(**) basis set. Relative to the regular MP2 results that could be afforded, the FMO2-MP2 error in the correlation energy did not exceed 0.003 a.u., the error in the correlation energy gradient did not exceed 0.000 05 a.u./bohr and the error in the correlation contribution to dipole moment did not exceed 0.03 debye. An approximation reducing computational load based on fragment separation was introduced and tested. The FMO2-MP2 method demonstrated nearly linear scaling and drastically reduced the memory requirements of the regular MP2, making possible calculations with several thousands basis functions using small Pentium clusters. As an example, (H(2)O)(64) with the 6-31 + + G(**) basis set (1920 basis functions) can be run in 1 Gbyte RAM and it took 136 s on a 40-node Pentium4 cluster.     

Calculation of intermolecular interactions in the benzene dimer using coupled-cluster and local electron correlation methods

Potential energy curves for the parallel-displaced, T-shaped and sandwich structures of the benzene dimer are computed with density fitted local second-order M?ller-Plesset perturbation theory (DF-LMP2) as well as with the spin-component scaled (SCS) variant of DF-LMP2. While DF-LMP2 strongly overestimates the dispersion interaction, in common with canonical MP2, the DF-SCS-LMP2 interaction energies are in excellent agreement with the best available literature values along the entire potential energy curves. The DF-SCS-LMP2 dissociation energies for the three structures are also compared with new complete basis set estimates of the interaction energies obtained from accurate coupled cluster (CCSD(T)) and DF-SCS-MP2 calculations. Since LMP2 is essentially free of basis set superposition errors, counterpoise corrections are not required. As a result, DF-SCS-LMP2 is computationally inexpensive and represents an attractive method for the study of larger pi-stacked systems such as truncated sections of DNA.     

Analytic gradient for second order Møller-Plesset perturbation theory with the polarizable continuum model based on the fragment molecular orbital method

A new energy expression is proposed for the fragment molecular orbital method interfaced with the polarizable continuum model (FMO/PCM). The solvation free energy is shown to be more accurate on a set of representative polypeptides with neutral and charged residues, in comparison to the original formulation at the same level of the many-body expansion of the electrostatic potential determining the apparent surface charges. The analytic first derivative of the energy with respect to nuclear coordinates is formulated at the second-order M?ller-Plesset (MP2) perturbation theory level combined with PCM, for which we derived coupled perturbed Hartree-Fock equations. The accuracy of the analytic gradient is demonstrated on test calculations in comparison to numeric gradient. Geometry optimization of the small Trp-cage protein (PDB: 1L2Y) is performed with FMO/PCM/6-31(+)G(d) at the MP2 and restricted Hartree-Fock with empirical dispersion (RHF/D). The root mean square deviations between the FMO optimized and NMR experimental structure are found to be 0.414 and 0.426 A? for RHF/D and MP2, respectively. The details of the hydrogen bond network in the Trp-cage protein are revealed.     

Performance of dispersion-corrected density functional theory for the interactions in ionic liquids

Potential energy curves for the dissociation of cation-anion associates representing the building units of ionic liquids have been computed with dispersion corrected DFT methods. Non-local van der Waals density functionals (DFT-NL) for the first time as well as an atom pair-wise correction method (DFT-D3) have been tested. Reference data have been computed at the extrapolated MP2/CBS and estimated CCSD(T)/CBS levels of theory. The investigated systems are combined from two cations (1-butyl-3-methylimidazolium and tributyl(methyl)posphonium) and three anions (chloride, dicyanamide, acetate). We find substantial stabilization from London dispersion energy near equilibrium of 5-7 kcal mol(-1) (about 5-6% of the interaction energy). Equilibrium distances are shortened by 0.03-0.09 ? and fundamental (inter-fragment) vibrational frequencies (which are in the range 140-180 cm(-1)) are increased by typically 10 cm(-1) when dispersion corrections are made. The dispersion-corrected hybrid functional potentials are in general in excellent agreement with the corresponding CCSD(T) reference data (typical deviations of about 1-2%). The DFT-D3 method performs unexpectedly well presumably because of cancellation of errors between the dispersion coefficients of the cations and anions. Due to self-interaction error, semi-local density functionals exhibit severe SCF convergence problems, and provide artificial charge-transfer and inaccurate interaction energies for larger inter-fragment distances. Although these problems may be alleviated in condensed phase simulations by effective Coulomb screening, only dispersion-corrected hybrid functionals with larger amounts of Fock-exchange can in general be recommended for such ionic systems.     

Role of the key mutation in the selective binding of avian and human influenza hemagglutinin to sialosides revealed by quantum-mechanical calculations

The selective binding between avian and human influenza A viral hemagglutinins (HA) subtype H3 and Neu5Acα2-3 and α2-6Gal (avian α2-3, human α2-6) is qualitatively rationalized by the fragment molecular orbital (FMO) method. We suggest a general model of analyzing protein-ligand interactions based on the electrostatic, polarization, dispersion, and desolvation components obtained from quantum-mechanical calculations at the MP2/6-31G(d) level with the polarizable continuum model of solvation. The favorable avian H3 (A/duck/Ukraine/1963)-avian α2-3 binding arises from the hydrophilic interaction between Gal-4 OH and side-chain NH(2)CO on Gln226, which is supported by the intermolecular hydrogen-bond network to the 1-COO group on Neu5Ac moiety. A substitution of Gln226Leu in the avian H3 HA1 domain increases the binding affinity to human α2-6 due to the Leu226···human α2-6 dispersion with a small entropic penalty during the complex formation. The remarkable human H3 (A/Aichi/2/1968)-human α2-6 binding is not governed by the Ser228-OH···OH-9 Neu5Ac hydrogen bond. These fragment-based chemical aspects can help design monovalent inhibitors of the influenza viral HA-sialoside binding and the simulation studies on the viral HAs-human α2-6 binding.     

Antigen�Cantibody interactions of influenza virus hemagglutinin revealed by the fragment molecular orbital calculation

Effective interactions between amino acid residues in antigen?Cantibody complex of influenza virus hemagglutinin (HA) protein can be evaluated in terms of the inter-fragment interaction energy (IFIE) analysis with the fragment molecular orbital (FMO) method, in which each fragment contains the side chain of corresponding amino acid residue. We have carried out the FMO-MP2 (second-order Moeller?CPlesset) calculation for the complex of HA antigen and Fab antibody of influenza virus H3N2 A/Aichi/2/68 and obtained the IFIE values between each amino acid residue in HA and the whole antibody as the sums over the residues contained in the latter. Combining this IFIE data with experimental data for hemadsorption activity of HA mutants, we succeeded in theoretically explaining the mutations in HA observed after the emergence of influenza virus H3N2 A/Aichi/2/68 in an earlier study, except for those of THR83. In the present study, we employ an alternative way of fragment division in the FMO calculation at the carbonyl C site of the peptide bond instead of the C_?? site used in the previous work, which provides revised IFIE values consistent with all the historical mutation data in the antigenic region E of HA including the case of THR83 in the present prediction scheme for probable mutations in HA.     

Ab initio quantum chemical calculation of electron density,electrostatic potential,and electric field of biomolecule based on fragment molecular orbital method

Efficient quantum chemical calculations of electrostatic properties, namely, the electron density (EDN), electrostatic potential (ESP), and electric field (EFL), were performed using the fragment molecular orbital (FMO) method. The numerical errors associated with the FMO scheme were examined at the HF, MP2, and RI‐MP2 levels of theory using 4 small peptides. As a result, the FMO errors in the EDN, ESP, and EFL were significantly smaller than the magnitude of the electron correlation effects, which indicated that the FMO method provides sufficiently accurate values of electrostatic properties. In addition, an attempt to reduce the computational effort was proposed by combining the FMO scheme and a point charge approximation. The error due to this approximation was examined using 2 proteins, prion protein and human immunodeficiency virus type 1 protease. As illustrative examples, the ESP values at the molecular surface of these proteins were calculated at the MP2 level of theory.     

Assessment of the performance of MP2 and MP2 variants for the treatment of noncovalent interactions

For many years, MP2 served as the principal method for the treatment of noncovalent interactions. Until recently, this was the only technique that could be used to produce reasonably accurate binding energies, with binding energy errors generally below ~35%, at a reasonable computational cost. The past decade has seen the development of many new methods with improved performance for noncovalent interactions, several of which are based on MP2. Here, we assess the performance of MP2, LMP2, MP2-F12, and LMP2-F12, as well as spin component scaled variants (SCS) of these methods, in terms of their abilities to produce accurate interaction energies for binding motifs commonly found in organic and biomolecular systems. Reference data from the newly developed S66 database of interaction energies are used for this assessment, and a further set of 38 complexes is used as a test set for SCS methods developed herein. The strongly basis set-dependent nature of MP2 is confirmed in this study, with the SCS technique greatly reducing this behavior. It is found in this work that the spin component scaling technique can effectively be used to dramatically improve the performance of MP2 and MP2 variants, with overall errors being reduced by factors of about 1.5-2. SCS versions of all MP2 variants tested here are shown to give similarly accurate overall results.     

Application of fragment molecular orbital scheme to silicon-containing systems

The fragment molecular orbital (FMO) scheme has been successfully used for a variety of large-scale molecules such as proteins and nucleic acids so far. We have applied the FMO calculations to the silicon-containing systems like polysilanes. The error caused by the fragmentation was examined by the Hartree–Fock method and the second-order Møller-Plesset (MP2) perturbation method for the ground state energy. The dynamic polarizability as a linear response property was also evaluated with and without the fragmentation. A series of numerical comparisons showed that the FMO scheme is applicable to silicon-based molecules with reasonable accuracy. This implied a potential availability of FMO calculations for the issues relevant to nanoscience and nanotechnology.     

Change in a protein's electronic structure induced by an explicit solvent: an ab initio fragment molecular orbital study of ubiquitin

The effect of solvation on the electronic structure of the ubiquitin protein was analyzed using the ab initio fragment molecular orbital (FMO) method. FMO calculations were performed for the protein in vacuo, and the protein was immersed in an explicit solvent shell as thick as 12 A at the HF or MP2 level by using the 6-31G* basis set. The protein's physical properties examined were the net charge, the dipole moment, the internal energy, and the solvent interaction energy. Comparison of the computational results revealed the following changes in the protein upon solvation. First, the positively charged amino acid residues on the protein surface drew electrons from the solvent, while the negatively charged ones transfer electrons to the solvent. Second, the dipole moment of the protein was enhanced as a result of the polarization. Third, the internal energy of the protein was destabilized, but the destabilization was more than compensated for by the generation of a favorable protein-solvent interaction. Finally, the energetic changes were elicited both by the electron correlation effect of the first solvent shell and by the electrostatic effect of more distant solvent molecules. These findings were consistent with the picture of the solvated protein being a polarizable molecule dissolved in a dielectric media.     

Improved supermolecular second order Møller-Plesset intermolecular interaction energies using time-dependent density functional response theory

The supermolecular second order Moller-Plesset (MP2) intermolecular interaction energy is corrected by employing time-dependent density functional (TDDFT) response theory. This is done by replacing the uncoupled second order dispersion contribution contained in the supermolecular MP2 energy with the coupled dispersion energy obtained from the TDDFT approach. Preliminary results for the rare gas dimers He2, Ne2, and Ar2 and a few structures of the (HF)2 and (H2O)2 dimers show that the conventional MP2 interaction energies are considerably improved by this procedure if compared to coupled cluster singles doubles with perturbative triples [CCSD(T)] interaction energies. However, the quality of the interaction energies obtained in this way strongly depends on the exchange-correlation potential employed in the monomer calculations: It is shown that an exact exchange-only potential surprisingly often performs better than an asymptotically corrected hybrid exchange-correlation potential. Therefore the method proposed in this work is similar to the method by Cybulski and Lytle [J. Chem. Phys., 127, 141102 (2007)] which corrects the supermolecular MP2 energies with a scaled dispersion energy from time-dependent Hartree-Fock. The results in this work are also compared to the combination of density functional theory and intermolecular perturbation theory.     

Communication: variational many-body expansion: accounting for exchange repulsion, charge delocalization, and dispersion in the fragment-based explicit polarization method

A fragment-based variational many-body (VMB) expansion method is described to directly account for exchange repulsion, charge delocalization (charge transfer) and dispersion interactions in the explicit polarization (X-Pol) method. The present VMB/X-Pol approach differs from other fragment molecular orbital (FMO) techniques in two major aspects. First, the wave function for the monomeric system is variationally optimized using standard X-Pol method, as opposed to the iterative update procedure adopted in FMO. Second, the mutual polarizations in the dimeric terms are also variationally determined, whereas single-point energy calculations of the individual dimers embedded in a static monomer field are used in FMO. The second-order (two-body) VMB (VMB2) expansion method is illustrated on a series of water hexamer complexes and one decamer cluster, making use of Hartree-Fock theory, MP2, and the PBE1 and M06 density functionals to represent the monomer and dimer fragments. The computed binding energies are within 2 kcal/mol of the corresponding results from fully delocalized calculations. Energy decomposition analyses reveal specific dimeric contributions to exchange repulsion, charge delocalization, and dispersion. Since the wave functions for one-body and all two-body terms are variationally optimized in VMB2 and X-Pol, it is straightforward to obtain analytic gradient without the additional coupled-perturbed Hartree-Fock step. Thus, the method can be useful for molecular dynamics simulations.     

Analytic energy gradient for second-order Møller-Plesset perturbation theory based on the fragment molecular orbital method

The first derivative of the total energy with respect to nuclear coordinates (the energy gradient) in the fragment molecular orbital (FMO) method is applied to second order M?ller-Plesset perturbation theory (MP2), resulting in the analytic derivative of the correlation energy in the external self-consistent electrostatic field. The completely analytic energy gradient equations are formulated at the FMO-MP2 level. Both for molecular clusters (H(2)O)(64) and a system with fragmentation across covalent bonds, a capped alanine decamer, the analytic FMO-MP2 energy gradients with the electrostatic dimer approximation are shown to be complete and accurate by comparing them with the corresponding numeric gradients. The developed gradient is parallelized with the parallel efficiency of about 97% on 32 Pentium4 nodes connected by Gigabit Ethernet.     

Assessment and acceleration of binding energy calculations for protein–ligand complexes by the fragment molecular orbital method

In the field of drug discovery, it is important to accurately predict the binding affinities between target proteins and drug applicant molecules. Many of the computational methods available for evaluating binding affinities have adopted molecular mechanics‐based force fields, although they cannot fully describe protein–ligand interactions. A noteworthy computational method in development involves large‐scale electronic structure calculations. Fragment molecular orbital (FMO) method, which is one of such large‐scale calculation techniques, is applied in this study for calculating the binding energies between proteins and ligands. By testing the effects of specific FMO calculation conditions (including fragmentation size, basis sets, electron correlation, exchange‐correlation functionals, and solvation effects) on the binding energies of the FK506‐binding protein and 10 ligand complex molecule, we have found that the standard FMO calculation condition, FMO2‐MP2/6‐31G(d), is suitable for evaluating the protein–ligand interactions. The correlation coefficient between the binding energies calculated with this FMO calculation condition and experimental values is determined to be R = 0.77. Based on these results, we also propose a practical scheme for predicting binding affinities by combining the FMO method with the quantitative structure–activity relationship (QSAR) model. The results of this combined method can be directly compared with experimental binding affinities. The FMO and QSAR combined scheme shows a higher correlation with experimental data (R = 0.91). Furthermore, we propose an acceleration scheme for the binding energy calculations using a multilayer FMO method focusing on the protein–ligand interaction distance. Our acceleration scheme, which uses FMO2‐HF/STO‐3G:MP2/6‐31G(d) at R_int = 7.0 Å, reduces computational costs, while maintaining accuracy in the evaluation of binding energy. © 2015 Wiley Periodicals, Inc.