Multithreaded parallelization of the energy and analytic gradient in the fragment molecular orbital method

The fragment molecular orbital method in GAMESS is parallelized in a multithreaded OpenMP implementation combined with the MPI version of the two-level generalized distributed data interface. The energy and analytic gradient in gas phase and the polarizable continuum model of solvation are parallelized in this hybrid three-level scheme, achieving a large memory footprint reduction and a high parallel efficiency on Intel Xeon Phi processors. The parallel efficiency is demonstrated on the Stampede2 and Theta supercomputers using up to 2048 nodes (262 144 threads).     

A new spline approach for data fitting

This paper presents an extension of the Modified Spline Technique (MST) formulation for data fitting named power spline, which gives consistent results for sets of data of concave and convex curves. This method is based on a technique which couples an implicit formulation of the maximum likelihood principle to the spline method, making the method suitable to fit data when no physical model is available. The MST method proved to be better than to the spline method and the extended spline fit technique (EST), because it provided accurate results for the first and second derivatives in sets of data where the EST solution developed inaccuracies. The EST method is a formulation that couples Least Squares to the spline method. There are, however, some sets of data where the MST method would show inconsistent solutions for the first and second derivatives. The power spline method eliminates these problems for concave or convex curves. Another improvement on the method is a more flexible choice for the interval boundaries.¹     

Expressed Sequence Tag Analysis of the Dinoflagellate Lingulodinium polyedrum During Dark Phase¶

To collect information on gene expression during the dark period in the luminous dinoflagellate Lingulodinium polyedrum, normalized complementary DNA (cDNA) libraries were constructed from cells collected during the first hour of night phase in a 12:12 h light‐dark cycle. A total of 4324 5′‐end sequence tags were isolated. The sequences were grouped into 2111 independent expressed sequence tags (EST) from which 433 groups were established by similarity searches of the public nonredundant protein database. Homology analysis of the total sequences indicated that the luminous dinoflagellate is more similar to land plants and animals (vertebrates and invertebrates) than to prokaryotes or algae. We also isolated three bioluminescence‐related (luciferase and two luciferinbinding proteins [LBP]) and 37 photosynthesis‐related genes. Interestingly, two kinds of LBP genes occur in multiple copies in the genome, in contrast to the single luciferase gene. These cDNA clones and EST sequence data should provide a powerful resource for future genome‐wide functional analyses for uncharacterized genes.     

Molecular interactions between estrogen receptor and its ligand studied by the ab initio fragment molecular orbital method

The ab initio fragment molecular orbital calculations were performed for molecular interactions of the whole estrogen receptor (ER) ligand-binding domain with a natural ligand, 17beta-estradiol (EST). The interaction energies of the ligand at the residue level were calculated using HF and MP2 methods with several basis sets. The charge-transfer (CT) interactions were also analyzed based on configuration analysis for fragment interaction. Strong electrostatic interactions were observed between the EST and surrounding charged/polarized residues, Glu353, Arg394, His524, and Thr347. Weak electrostatic and significant van der Waals dispersion interactions were observed between the EST and the many surrounding hydrophobic residues. Together with the experimental interpretations, both interactions equally contributed to the total binding energies, and it was found that the inclusion of electron correlation was essential to obtain an appropriate picture of the interaction. The strongest interaction energy was observed between Glu353 and the EST, and the CT interactions from the lone-pair orbital of the carbonyl oxygen of Glu353 to the sigma(OmicronEta) orbital of the hydroxyl group of EST were found to be important. The CT interactions from the lone-pair orbital of EST to the sigma(NuEta) of Arg394 and from the lone-pair orbital of EST to the sigma(NuEta) of His524 were also observed. These CT interactions occurred through the hydrogen-bond networks between the ER and EST. Therefore, electron donations from the ER to the EST and electron back-donations from EST to the ER were characteristic of ER-ligand binding. Our approach provides a powerful tool to understanding detailed molecular interactions at the quantum mechanical level.     

Calculating the density of states for large‐scale molecular systems

We present a new method for calculating the density of states of large‐scale molecular systems. The new algorithm does not require calculation of all eigenvalues of the force constant matrix. The histogram for the eigenvalue distribution is constructed by performing a sequence of sparse LDL^T matrix factorizations. Moreover, the entire calculation can be fully parallelized on a multiprocessor computer system.     

Food-induced esterase electromorphs in Carinarion spp. and their effects on taxonomic data analysis (Gastropoda, Pulmonata, Arionidae)

Nonspecific esterases (EST) are often used to measure genetic variation, yet they may be influenced by environmental factors such as food, climate and age. This may produce misleading similarity indices and genetic diversity estimates (i.e., clone or strain diversities in uniparental organisms). Therefore, polyacrylamide gel electrophoresis (PAGE) and isoelectric focusing (IEF) were used to investigate environmental effects on the EST variation in natural Carinarion populations, as well as in 45 individuals that were raised individually on carrots to produce offspring by selfing. Food effects on EST profiles in these progenies were examined by raising them on different food items (lettuce, nettle, or paper). Our results indicated that: (i) Arion (Carinarion) fasciatus and A. (C.) silvaticus show species-specific EST profiles, (ii) A. fasciatus-like outcrossers most probably are conspecific with A. fasciatus s.s., (iii) not all EST variation has a Mendelian basis since lettuce and nettle altered EST profiles, and (iv) food effects on EST profiles differed strongly between individuals. Although food-induced EST profiles did not affect taxonomic interpretations, they did inflate genetic diversity estimates and thus provided misleading population-genetic data.     

PHI: A powerful new program for the analysis of anisotropic monomeric and exchange‐coupled polynuclear d‐ and f‐block complexes

A new program, PHI, with the ability to calculate the magnetic properties of large spin systems and complex orbitally degenerate systems, such as clusters of d‐block and f‐block ions, is presented. The program can intuitively fit experimental data from multiple sources, such as magnetic and spectroscopic data, simultaneously. PHI is extensively parallelized and can operate under the symmetric multiprocessing, single process multiple data, or GPU paradigms using a threaded, MPI or GPU model, respectively. For a given problem PHI is been shown to be almost 12 times faster than the well‐known program MAGPACK, limited only by available hardware. © 2013 Wiley Periodicals, Inc.     

A fast, scalable method for the parallel evaluation of distance-limited pairwise particle interactions

Classical molecular dynamics simulations of biological macromolecules in explicitly modeled solvent typically require the evaluation of interactions between all pairs of atoms separated by no more than some distance R, with more distant interactions handled using some less expensive method. Performing such simulations for periods on the order of a millisecond is likely to require the use of massive parallelism. The extent to which such simulations can be efficiently parallelized, however, has historically been limited by the time required for interprocessor communication. This article introduces a new method for the parallel evaluation of distance-limited pairwise particle interactions that significantly reduces the amount of data transferred between processors by comparison with traditional methods. Specifically, the amount of data transferred into and out of a given processor scales as O(R(3/2)p(-1/2)), where p is the number of processors, and with constant factors that should yield a substantial performance advantage in practice.     

A hierarchical clustering algorithm for MIMD architecture

Hierarchical clustering is the most often used method for grouping similar patterns of gene expression data. A fundamental problem with existing implementations of this clustering method is the inability to handle large data sets within a reasonable time and memory resources. We propose a parallelized algorithm of hierarchical clustering to solve this problem. Our implementation on a multiple instruction multiple data (MIMD) architecture shows considerable reduction in computational time and inter-node communication overhead, especially for large data sets. We use the standard message passing library, message passing interface (MPI) for any MIMD systems.     

Multi-state multi-mode nuclear dynamics on three isomers of C6H4F2+ using parallelized TDDVR approach

We have performed molecular dynamics on the three isomers of the difluorobenzene radical cation (C(6)H(4)F(2)(+)) after excitation from the ground state to a specific higher electronically excited state by using our recently implemented parallelized Time-Dependent Discrete Variable Representation (TDDVR) methodology. A five-state eleven-mode realistic model Hamiltonian for o-C(6)H(4)F(2)(+) and two separate five-state ten-mode Hamiltonians for m- and p-isomer of the same radical cation are considered, where those five electronic states are interconnected through several conical intersections in the vicinity of the Franck-Condon (FC) region and thus the dynamics for each case become complex. The photoelectron, mass analyzed threshold ionization spectra and population profiles obtained by using our TDDVR approach show reasonably good agreement with the results obtained by multiconfiguration time dependent Hartree (MCTDH) method. It is worthwhile to mention that the parallelized TDDVR algorithm reduces the computation time by more than an order of magnitude compared to its serial analog and, therefore, such approach appears to be a good compromise between accuracy and speed for a large molecular system.     

Effect of ions on the polymorphism, effective charge, and stability of human telomeric DNA. Photon correlation spectroscopy and circular dichroism studies

The effect of different ions on the formation and behavior of quadruplex structures of the human telomere sequence d(TTAGGG)(4) has been studied by photon correlation spectroscopy (PCS) and circular dichroism (CD). The saturation and melting curves obtained in the presence of K(+), Na(+), Rb(+), Li(+), Cs(+), and Sr(2+) ions were recorded by CD spectroscopy and indicated the formation of monomeric quadruplexes. Analysis of the saturation curves obtained at 2 degrees C has shown that the presence of a single Sr(2+) ion per oligomer is sufficient for the formation of a monomeric quadruplex of the DNA sequence studied. In the presence of SrCl(2) at a concentration of 50 mM, the formation of tetrameric quadruplexes has been detected. The effect of Sr(2+) ions on the formation of quadruplex structures by the human telomere sequence d(TTAGGG)(4) is stronger and different from that of the other ions tested. The paper also presents results of a study of electrostatic interactions in solution. The translation diffusion coefficients D(T) of the structures present in solution have been determined by photon correlation spectroscopy and the effective charges on the structures have been calculated by combining the experimental data with the results based on the coupled mode theory. Analysis of the melting points monitored by the CD method has permitted a determination of Deltan, the number of ions released in the process of thermal denaturation. All the results are in good agreement with the predictions based on the theory of polyelectrolytes. The effect of ions on the formation and behavior of quadruplex structures of the human telomere sequence d(TTAGGG)(4) has been studied by photon correlation spectroscopy and circular dichroism.     

Exploring chemistry with the fragment molecular orbital method

The fragment molecular orbital (FMO) method makes possible nearly linear scaling calculations of large molecular systems, such as water clusters, proteins and DNA. In particular, FMO has been widely used in biochemical applications involving protein-ligand binding and drug design. The method has been efficiently parallelized suitable for petascale computing. Many commonly used wave functions and solvent models have been interfaced with FMO. We review the historical background of FMO, and summarize its method development and applications.     

Parallelized integral-direct CIS(D) calculations with multilayer fragment molecular orbital scheme

We have developed a parallelized integral-direct code of the perturbative doubles correction for configuration interaction with singles, proposed as CIS(D) by Head-Gordon et al. (Chem Phys Lett 219:21, 1994). The CIS(D) method provides the energy corrections both of the relaxation and differential correlation for the respective CIS excited states. The implementation of CIS(D) is based on our original algorithm for the second-order Møller–Plesset perturbation (MP2) calculations (Mochizuki et al. in Theor Chem Acc 112:442, 2004). There is no need to communicate bulky intermediate data among worker processes of the parallelized execution. This CIS(D) code is then incorporated into a developer version of ABINIT-MP program, in order to improve the overestimation in excitation energies calculated by the CIS method in conjunction with the multilayer fragment molecular orbital scheme (MLFMO-CIS) (Mochizuki et al. in Chem Phys Lett 406:283, 2005). The MLFMO-CIS(D) method is first used in evaluating the lowest n\(\pi^{*}\) excitation energy of the hydrated formaldehyde. The photoactive yellow protein (PYP) is the second target of MLFMO-CIS(D) calculation. Through these applications, it is shown that the CIS(D) correction improves the CIS results favorably.     

High-performance transformation of protein structure representation from internal to Cartesian coordinates

We present a highly parallel algorithm to convert internal coordinates of a polymeric molecule into Cartesian coordinates. Traditionally, converting the structures of polymers (e.g., proteins) from internal to Cartesian coordinates has been performed serially, due to an inherent linear dependency along the polymer chain. We show this dependency can be removed using a tree-based concatenation of coordinate transforms between segments, and then parallelized efficiently on graphics processing units (GPUs). The conversion algorithm is applicable to protein engineering and fitting protein structures to experimental data, and we observe an order of magnitude speedup using parallel processing on a GPU compared to serial execution on a CPU.     

The proteome of maize leaves: use of gene sequences and expressed sequence tag data for identification of proteins with peptide mass fingerprints.

As a first step in establishing a proteome database for maize, we have embarked on the identification of the leaf proteins resolved on two-dimensional (2-D) gels. We detected nearly 900 spots on the gels with a pH 4-7 gradient and over 200 spots on the gels with a pH 6-11 gradient when the proteins were visualized with colloidal Coomassie blue. Peptide mass fingerprints for 300 protein spots were obtained with matrix assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometer and 149 protein spots were identified using the protein databases. We also searched the pdbEST databases to identify the leaf proteins and verified 66% of the protein spots that had been identified using the protein databases. Sixty-seven additional protein spots were identified from expressed sequence tags (ESTs). Many abundant leaf proteins are present in multiple spots. Functions of over 50% of the abundant leaf proteins are either unknown or hypothetical. Our results show that EST databases in conjunction with peptide mass fingerprints can be used for identifying proteins from organisms with incomplete genome sequence information.     

Some aspects of parallelization of two-electron integrals in molecular orbital programs

Evaluation of two-electron integrals forms a substantial part of the CPU time for any ab initio molecular orbital program. This part of the package, “MICROMOL”, is parallelized. However, this parallelization leads to only sublinear speedups (typically 3 on a 4-node machine). In view of these results, the task of development of an efficient program for two-electron integrals suitable for the parallel environment has been taken up. The program is written in FORTRAN considering specific symmetry features and application of rigorous bounds. This program is further parallelized with a good load balancing strategy. The molecules used as the test cases are: trans-butadiene, benzene, nitrobenzene, naphtalene and cytosine, with 3G and 4–31G basis sets. The results indicate that the parallel version of this program gives a typical speedup of 3.6 for a 3G basis set and approximately 3.4 for a 4–31G basis set for all the molecules tested. The sequential version of this program is ～1.2 times faster than the sequential version of MICROMOL, whereas the parallel version is ～1.4 times faster than the parallelized MICROMOL.     

Newly identified [correction of dentified] members of the TNF receptor superfamily (mTNFRH1 and mTNFRH2) inhibit T-cell proliferation

By searching an EST database, we identified two TNF receptor superfamily members (named mTNFRH1 and mTNFRH2). Amino acid sequences are highly conserved between the two receptors (78% identity). The chromosomal loci of mTnfrh1 and mTnfrh2 genes are found in distal chromosome 7 in the mouse. mTNFRH1 and mTNFRH2 do not contain the cytoplasmic domain, indicating that they might function as decoy receptors. Furthermore, an alternatively spliced form of mTNFRH1 was found which contains neither the transmembrane domain nor the cytoplasmic domain, thus presumably existing as a soluble form. Northern blot analysis showed that mTnfrh1 mRNA was negligibly expressed in tissues, while mTnfrh2 mRNA was strongly expressed in spleen, lung, liver, kidney, and testis. When the extracellular domains of mTNFRH1 and mTNFRH2 were expressed in bacteria, their molecular weight of extracellular region was approximately 15 kDa. Both of the soluble forms were effective in inhibiting T-cell proliferation stimulated by anti-CD3 monoclonal antibody. Our data suggest that mTNFRH1 and mTNFRH2 may be implicated in exerting a modulatory role in the immune response.     

Creating hierarchical models of protein families based on Expressed Sequence Tags: the "Sprockets" analysis pipeline

We have created an analysis pipeline called Sprockets, which can be used to classify proteins into various hierarchical “families”, and build searchable models of these families. The construction of these families is based on data from Expressed Sequence Tags (ESTs) and Coding DNA Sequences (CDSs), making Sprockets clusters especially suitable for studying gene families in organisms for which the completely sequenced genome does not (yet) exist. The pipeline consists of two main parts: pair-wise analysis and grouping of sequences with Z-score statistics, followed by hierarchical splitting of clusters into alignable protein families. Various computational and statistical techniques applied in Sprockets allow it to act like a massive and selective multiple sequence alignment engine for combining individual sequence collections and related public sequences. The end result is a database of gene Hidden Markov Models, each related to the other by three levels of similarity: secondary structure, function and evolutionary origin. For a sample 20,000 EST set from Lactuca spp., Sprockets provided a 9% improvement in mapping of function to unknown sequences over traditional pair-wise search methods and InterPro mapping.     

Diffuse component of solar ultraviolet radiation in tree shade

The first set of quantitative data of diffuse erythemal UV and UV-A radiation in tree shade at a sub-tropical Southern Hemisphere latitude is presented. Over the summer, approximately 60% of the erythemal UV radiation in tree shade is due to the diffuse component. Similarly, approximately 56% of the UV-A radiation in tree shade is due to the diffuse component. In tree shade these diffuse UV percentages are relatively constant from the morning to noon to afternoon periods. In comparison, in full sun, there is a decrease in the percentage of diffuse UV from morning to noon to afternoon. The exposures to diffuse UV on a horizontal plane in tree shade between 9:00 EST and 15:00 EST are of the order of 4 MED (minimum erythemal dose) and 14 J cm(-2) for erythemal UV and UV-A, respectively. The high diffuse UV component in the shade may result in high UV exposures not only to unprotected parts of the body on a horizontal plane, but also in equally high UV irradiances to parts of the body, including the eyes and face, that are not UV protected.