Papers by Magnus Lundborg
The introduction of accelerator devices such as graphics processing units (GPUs) has had profound... more The introduction of accelerator devices such as graphics processing units (GPUs) has had profound impact on molecular dynamics simulations and has enabled order-of-magnitude performance advances using commodity hardware. To fully reap these benefits, it has been necessary to reformulate some of the most fundamental algorithms, including the Verlet list, pair searching and cut-offs. Here, we present the heterogeneous parallelization and acceleration design of molecular dynamics implemented in the GROMACS codebase over the last decade. The setup involves a general cluster-based approach to pair lists and non-bonded pair interactions that utilizes both GPUs and CPU SIMD acceleration efficiently, including the ability to load-balance tasks between CPUs and GPUs. The algorithm work efficiency is tuned for each type of hardware, and to use accelerators more efficiently we introduce dual pair lists with rolling pruning updates. Combined with new direct GPU-GPU communication as well as GPU ...
The Journal of Chemical Physics
The introduction of accelerator devices such as graphics processing units (GPUs) has had profound... more The introduction of accelerator devices such as graphics processing units (GPUs) has had profound impact on molecular dynamics simulations and has enabled order-of-magnitude performance advances using commodity hardware. To fully reap these benefits, it has been necessary to reformulate some of the most fundamental algorithms, including the Verlet list, pair searching, and cutoffs. Here, we present the heterogeneous parallelization and acceleration design of molecular dynamics implemented in the GROMACS codebase over the last decade. The setup involves a general cluster-based approach to pair lists and non-bonded pair interactions that utilizes both GPU and central processing unit (CPU) single instruction, multiple data acceleration efficiently, including the ability to load-balance tasks between CPUs and GPUs. The algorithm work efficiency is tuned for each type of hardware, and to use accelerators more efficiently, we introduce dual pair lists with rolling pruning updates. Combined with new direct GPU-GPU communication and GPU integration, this enables excellent performance from single GPU simulations through strong scaling across multiple GPUs and efficient multi-node parallelization.
Biophysical Journal
The stratum corneum is the outermost layer of human skin and the primary barrier toward the envir... more The stratum corneum is the outermost layer of human skin and the primary barrier toward the environment. The barrier function is maintained by stacked layers of saturated long-chain ceramides, free fatty acids, and cholesterol. This structure is formed through a reorganization of glycosylceramide-based bilayers with cubic-like symmetry into ceramide-based bilayers with stacked lamellar symmetry. The process is accompanied by deglycosylation of glycosylceramides and dehydration of the skin barrier lipid structure. Using coarse-grained molecular dynamics simulation, we show the effects of deglycosylation and dehydration on bilayers of human skin glycosylceramides and ceramides, folded in three dimensions with cubic (gyroid) symmetry. Deglycosylation of glycosylceramides destabilizes the cubic lipid bilayer phase and triggers a cubic-to-lamellar phase transition. Furthermore, subsequent dehydration of the deglycosylated lamellar ceramide system closes the remaining pores between adjacent lipid layers and locally induces a ceramide chain transformation from a hairpin-like to a splayed conformation.
Copernicus, a hybrid dataflow and peer-to-peer scientific computing platform for efficient large-scale ensemble sampling
Future Generation Computer Systems
Compute-intensive applications have gradually changed focus from massively parallel supercomputer... more Compute-intensive applications have gradually changed focus from massively parallel supercomputers to capacity as a resource obtained on-demand. This is particularly true for the large-scale adoption of cloud computing and MapReduce in industry, while it has been difficult for traditional high-performance computing (HPC) usage in scientific and engineering computing to exploit this type of resources. However, with the strong trend of increasing parallelism rather than faster processors, a growing number of applications target parallelism already on the algorithm level with loosely coupled approaches based on sampling and ensembles. While these cannot trivially be formulated as MapReduce, they are highly amenable to throughput computing. There are many general and powerful frameworks, but in particular for sampling-based algorithms in scientific computing there are some clear advantages from having a platform and scheduler that are highly aware of the underlying physical problem. Here, we present how these challenges are addressed with combinations of dataflow programming, peer-to-peer techniques and peer-to-peer networks in the Copernicus platform. This allows automation of sampling-focused workflows, task generation, dependency tracking, and not least distributing these to a diverse set of compute resources ranging from supercomputers to clouds and distributed computing (across firewalls and fragile networks). Workflows are defined from modules using existing programs, which makes them reusable without programming requirements. The system achieves resiliency by handling node failures transparently with minimal loss of computing time due to checkpointing, and a single server can manage hundreds of thousands of cores e.g. for computational chemistry applications. Hybrid dataflow and peer-to-peer computing to fully automated ensemble sampling.The platform automatically distributes workloads and manages them resilientlyProblems are defined as workflow by reusing existing software and scripts.Portability in networks where parts are behind firewalls.
Cell Reports
Highlights d Clathrin-independent endocytosis occurs in response to a decrease in cell volume d R... more Highlights d Clathrin-independent endocytosis occurs in response to a decrease in cell volume d Reduced surface tension induces assembly of GRAF1 via the BAR and PH domains d GRAF1 activity suppresses cellular blebbing d Cancer cells lacking GRAF1 are more invasive
Fragment-based design of norovirus inhibitors by molecular docking and LIE binding free energy calculations
Fragment-based design of norovirus inhibitors by molecular docking and LIE binding free energy ca... more Fragment-based design of norovirus inhibitors by molecular docking and LIE binding free energy calculations
Journal of Chemical Theory and Computation, 2015
Computational chemistry and other simulation fields are critically dependent on computing resourc... more Computational chemistry and other simulation fields are critically dependent on computing resources, but few problems scale efficiently to the hundreds of thousands of processors available in current supercomputersparticularly for molecular dynamics. This has turned into a bottleneck as new hardware generations primarily provide more processing units rather than making individual units much faster, which simulation applications are addressing by increasingly focusing on sampling with algorithms such as free-energy perturbation, Markov state modeling, metadynamics, or milestoning. All these rely on combining results from multiple simulations into a single observation. They are potentially powerful approaches that aim to predict experimental observables directly, but this comes at the expense of added complexity in selecting sampling strategies and keeping track of dozens to thousands of simulations and their dependencies. Here, we describe how the distributed execution framework Copernicus allows the expression of such algorithms in generic workf lows: dataflow programs. Because dataflow algorithms explicitly state dependencies of each constituent part, algorithms only need to be described on conceptual level, after which the execution is maximally parallel. The fully automated execution facilitates the optimization of these algorithms with adaptive sampling, where undersampled regions are automatically detected and targeted without user intervention. We show how several such algorithms can be formulated for computational chemistry problems, and how they are executed efficiently with many loosely coupled simulations using either distributed or parallel resources with Copernicus.
Carbohydrate Structural Studies. Towards Automatic Structure Determination
The journal of physical chemistry. B, Jan 22, 2015
Free energy calculation has long been an important goal for molecular dynamics simulation and for... more Free energy calculation has long been an important goal for molecular dynamics simulation and force field development, but historically it has been challenged by limited performance, accuracy, and creation of topologies for arbitrary small molecules. This has made it difficult to systematically compare different sets of parameters to improve existing force fields, but in the past few years several authors have developed increasingly automated procedures to generate parameters for force fields such as Amber, CHARMM, and OPLS. Here, we present a new framework that enables fully automated generation of GROMACS topologies for any of these force fields and an automated setup for parallel adaptive optimization of high-throughput free energy calculation by adjusting lambda point placement on the fly. As a small example of this automated pipeline, we have calculated solvation free energies of 50 different small molecules using the GAFF, OPLS-AA, and CGenFF force fields and four different wa...
Structural studies of the O-antigen polysaccharide from Escherichia coli TD2158 having O18 serogroup specificity and aspects of its interaction with the tailspike endoglycosidase of the infecting bacteriophage HK620
Carbohydrate Research, 2012
We have analyzed the O-antigen polysaccharide of the previously uncharacterized Escherichia coli ... more We have analyzed the O-antigen polysaccharide of the previously uncharacterized Escherichia coli strain TD2158 which is a host of bacteriophage HK620. This bacteriophage recognizes and cleaves the polysaccharide with its tailspike protein (TSP). The polysaccharide preparation as well as oligosaccharides obtained from HK620TSP endoglycosidase digests were analyzed with NMR spectroscopy. Additionally, sugar analysis was performed on the O-antigen polysaccharide and MALDI-TOF MS was used in oligosaccharide analysis. The present study revealed a heterogeneous polysaccharide with a hexasaccharide repeating unit of the following structure: α-D-Glcp-(1→6|) →2)-α-L-Rhap-91→6)-α-D-Glcp-(1→4)-α-D-Ga|lp-(1→3)-α-D-GlcpNAc-(1→ β-D-Glcp/β-D-GlcpNAc-(1→3) A repeating unit with a D-GlcNAc substitution of D-Gal has been described earlier as characteristic for serogroup O18A1. Accordingly, we termed repeating units with D-Glc substitution at D-Gal as O18A2. NMR analyses of the polysaccharide confirmed that O18A1- and O18A2-type repeats were present in a 1:1 ratio. However, HK620TSP preferentially bound the D-GlcNAc-substituted O18A1-type repeating units in its high affinity binding pocket with a dissociation constant of 140 μM and disfavored the O18A2-type having a β-D-Glcp-(1→3)-linked group. As a result, in hexasaccharide preparations, O18A1 and O18A2 repeats were present in a 9:1 ratio stressing the clear preference of O18A1-type repeats to be cleaved by HK620TSP.
Carbohydrate Research, 2013
This is the accepted version of a paper published in Carbohydrate Research. This paper has been p... more This is the accepted version of a paper published in Carbohydrate Research. This paper has been peer-reviewed but does not include the final publisher proof-corrections or journal pagination.
Carbohydrates are abundant in nature and have functions, ranging from energy storage to acting as... more Carbohydrates are abundant in nature and have functions, ranging from energy storage to acting as structural components. Analysis of carbohydrate structures is important and can be used for, for instance, clinical diagnosis of diseases as well as in bacterial studies. The complexity of glycans makes it difficult to determine their structures. NMR spectroscopy is an advanced method that can be used to examine carbohydrates at the atomic level, but full assignments of the signals require much work. Reliable automation of this process would be of great help. Herein studies of Escherichia coli O-antigen polysaccharides are presented, both a structure determination by NMR and also research on glycosyltransferases which assemble the polysaccharides. The computer program CASPER has been improved to assist in carbohydrate studies and in the long run make it possible to automatically determine structures based only on NMR data. Detailed computer studies of glycans can shed light on their interactions with proteins and help find inhibitors to prevent unwanted binding. The WaaG glycosyltransferase is important for the formation of E. coli lipopolysaccharides. Molecular docking analyses of structures confirmed to bind this enzyme have provided information on how inhibitors could be composed. Noroviruses cause gastroenteritis, such as the winter vomiting disease, after binding human histo-blood group antigens. In one of the projects, fragment-based docking, followed by molecular dynamics simulations and binding free energy calculations, was used to find competitive binders to the P domain of the capsid of the norovirus VA387. These novel structures have high affinity and are a very good starting point for developing drugs against noroviruses. The protein targets in these two projects are carbohydrate binding, but the techniques are general and can be applied to other research projects.
Structural studies of the O-antigenic polysaccharides from the enteroaggregative Escherichia coli strain 94/D4 and the international type strain Escherichia coli O82
ECODAB (the E. coli O-antigen database) has been expanded to include information about glycosyltr... more ECODAB (the E. coli O-antigen database) has been expanded to include information about glycosyltransferases (GTs) involved in the assembly of the O-antigen polysaccharide. Similarity searches have been performed to be able to determine GT functions that have not been reported prior to this work. In addition to suggesting the function of 179 GTs, the approach leads to the prediction of part of the O-antigen structures of a number of serogroups. The procedure suggests a novel way of combining genetic information with experimental techniques in structural analysis of oligoand polysaccharides.
Structural Analysis of Glycans by NMR Chemical Shift Prediction
Structural determination of N- and O-linked glycans as well as polysaccharides is hampered by the... more Structural determination of N- and O-linked glycans as well as polysaccharides is hampered by the limited spectral dispersion. The computerized approach CASPER, an acronym for computer assisted spectrum evaluation of regular polysaccharides, uses liquid state NMR data to elucidate carbohydrate structure based on agreement with predicted (1)H and (13)C chemical shifts. We here demonstrate developments based on multiple through-bond J-based correlations that significantly enhance the credence to the sequence connectivities proposed in the analysis exemplified by an oligosaccharide and a bacterial polysaccharide. The approach is also suitable for predicting (1)H and (13)C NMR chemical shifts of synthesized oligosaccharides and glycoconjugates, thereby corroborating a proposed structure.
We have investigated the influence of post-filtering virtual screening results, with pharmacophor... more We have investigated the influence of post-filtering virtual screening results, with pharmacophoric features generated from an X-ray structure, on enrichment rates. This was performed using three docking softwares, zdock+, Surflex and FRED, as virtual screening tools and pharmacophores generated in UNITY from co-crystallized complexes. Sets of known actives along with 9997 pharmaceutically relevant decoy compounds were docked against six chemically diverse protein targets namely CDK2, COX2, ERa, fXa, MMP3, and NA. To try to overcome the inherent limitations of the well-known docking problem, we generated multiple poses for each compound. The compounds were first ranked according to their scores alone and enrichment rates were calculated using only the top scoring pose of each compound. Subsequently, all poses for each compound were passed through the different pharmacophores generated from co-crystallized complexes and the enrichment factors were recalculated based on the top-scoring passing pose of each compound. Post-filtering with a pharmacophore generated from only one X-ray complex was shown to increase enrichment rates in all investigated targets compared to docking alone. This indicates that this is a general method, which works for diverse targets and different docking softwares.
EUROCarbDB: An open-access platform for glycoinformatics
Complete 1H and 13C NMR chemical shift assignments of mono-, di-, and trisaccharides as basis for NMR chemical shift predictions of polysaccharides using the computer program casper
Carbohydrate Research, 2011
The computer program casper uses 1H and 13C NMR chemical shift data of mono- to trisaccharides fo... more The computer program casper uses 1H and 13C NMR chemical shift data of mono- to trisaccharides for the prediction of chemical shifts of oligo- and polysaccharides. In order to improve the quality of these predictions the 1H and 13C, as well as 31P when applicable, NMR chemical shifts of 30 mono-, di-, and trisaccharides were assigned. The reducing sugars gave
CarbBuilder: an adjustable tool for building 3D molecular structures of carbohydrates for molecular simulation
E-Science (e-Science), …, 2011
CarbBuilder is a software tool for building 3D structures of carbohydrates, which are the most st... more CarbBuilder is a software tool for building 3D structures of carbohydrates, which are the most structurally varied of all molecular classes. CarbBuilder was designed with the dual aims of portability and adaptability, using an iterative software development approach. CarbBuilder employs a simple algorithm, using heuristics based upon experimental data to convert a primary structure description of a carbohydrate molecule into a three-dimensional structure file. This straightforward approach means that CarbBuilder can be easily ...
An efficient and extensible format, library, and API for binary trajectory data from molecular simulations
Molecular dynamics simulations is an important application in theoretical chemistry, and with the... more Molecular dynamics simulations is an important application in theoretical chemistry, and with the large high-performance computing resources available today the programs also generate huge amounts of output data. In particular in life sciences, with complex biomolecules such as proteins, simulation projects regularly deal with several terabytes of data. Apart from the need for more cost-efficient storage, it is increasingly important to be able to archive data, secure the integrity against disk or file transfer errors, to provide rapid access, and facilitate exchange of data through open interfaces. There is already a whole range of different formats used, but few if any of them (including our previous ones) fulfill all these goals. To address these shortcomings, we present "Trajectory Next Generation" (TNG)--a flexible but highly optimized and efficient file format designed with interoperability in mind. TNG both provides state-of-the-art multiframe compression as well as a container framework that will make it possible to extend it with new compression algorithms without modifications in programs using it. TNG will be the new file format in the next major release of the GROMACS package, but it has been implemented as a separate library and API with liberal licensing to enable wide adoption both in academic and commercial codes.
Uploads
Papers by Magnus Lundborg