Log in

EPSRC High End Computing Consortia Mid-term Review, January 2015

 

Adrian Mulholland (AJM) (Chair) - Professor of Chemistry, University of Bristol
Phil Biggin (PCB) - Associate Professor of Computational Biochemistry, University of Oxford
Jonathan Essex (JWE) - Professor of Chemistry, University of Southampton
Francesco Gervasio (FLG) - Chair of Biomolecular Modelling, UCL; Thomas Young Centre
Sarah Harris (SAH) - Lecturer in Biological Physics, School of Physics and Astronomy, University of Leeds and a member of the Astbury Centre for Structural Molecular Biology
Richard Henchman (RHH) - Lecturer, School of Chemistry at the University of Manchester
David Huggins (DH) - MRC Fellow, Condensed Matter Group, University of Cambridge
Syma Khalid (SK) - Associate Professor in Chemistry, University of Southampton
Charles Laughton (CAL) - Associate Professor in Molecular Recognition, Pharmacy, Nottingham
Julien Michel (JM) - Royal Society University Research Fellow, University of Edinburgh
Edina Rosta (ER) - Lecturer in Computational Chemistry, King’s College London
Mark Sansom (MS) - David Phillips Professor of Molecular Biophysics, University of Oxford
Martyn Winn (MW) - Group Leader in Computational Biology, STFC Daresbury Laboratory


Introduction

Biomolecular simulations are now making significant contributions to a wide variety of problems in drug design and development, biocatalysis, bio and nano-technology, chemical biology and medicine. The UK has a strong and growing community in this field, recognized by the establishment in 2011 by EPSRC of CCP-BioSim (ccpbiosim.ac.uk), the UK Collaborative Computational Project for Biomolecular Simulation at the Life Sciences Interface, and its renewal in 2015. There is a clear, growing and demonstrable need for HEC in this field. HECBioSim, the UK HEC Biomolecular Simulation Consortium, was established in March 2013, and works closely with and complements CCP-BioSim.

Most of the members of the Consortium are experienced users of high end computing. Members of the Consortium have, e.g., served on the HECToR Resource Allocation Panel, including the current and previous Chairs of the RAP. The Consortium welcomes new members across the whole community. Since establishing the Consortium, several new members (FLG, DH, ER) have joined the Management Group. Many of the projects awarded HECToR and ARCHER time under the Consortium do not involve CCP-BioSim or HECBioSim Management Group members, demonstrating the openness of HECBioSim and its support of the biomolecular simulation community in the UK. A number of other researchers have expressed interest in joining and it is our expectation that other researchers will join HECBioSim in future. We actively engage with structural and chemical biologists and industrial researchers. We foster interactions between computational, experimental and industrial scientists (see e.g. case studies; members of the Consortium have excellent links with many pharmaceutical, chemical and biotechnology companies). HECBioSim also develops tools to help non-experts become proficient users of biomolecular simulation on HEC resources.

 

Scientific Area

Molecular simulations can provide molecular level understanding of how biological macromolecules function. E.g. molecular dynamics (MD) simulations have been instrumental in the growing recognition that protein dynamics are crucial for their biological function; understanding dynamics is essential for analysing activity. Biomolecular simulation and modelling is vital e.g. to the pharmaceutical industry, where it is an integral part of drug design and development. Longer term, biomolecular simulation has the potential to contribute to improvements in health and quality of life. Advances in hardware and software now allow simulations on the microsecond (and even millisecond) timescale, enabling direct links with experiments.

Simulations have proved crucial in analysing protein folding, mechanisms of biological catalysis, and how membrane proteins interact with lipid bilayers. Biomolecular simulations contribute to drug development (e.g. in structure-based drug design and predictions of metabolism) and design of biomimetic catalysts, and in understanding the molecular bases of disease and drug resistance. Cutting-edge biomolecular simulation demands HEC resources: e.g. large-scale simulations of ‘biological machines’ such as the ribosome, proton pumps and motors, membrane receptor complexes and even whole viruses. Different methods are applicable to different problems in biomolecular science, e.g, coarse-grained methods allow simulations on larger scales, and combined quantum mechanics/molecular mechanics (QM/MM) methods can model chemical mechanisms of enzyme catalysis. A particular challenge is the integration of different types of simulations across length and timescales.


Progress Report

 

Software Development

We aim to develop tools to assist large-scale simulation e.g. for tools to allow interoperability between different software packages, and the exchange of data and metadata and contribute to the development of data standards for biomolecular simulation. This will also facilitate analysis of the large volumes of data produced by simulations at large length- and timescales. There is a need for new tools to facilitate setup, simulation and analysis of large systems (e.g. FESetup). We also focus on optimization of existing tools. The tools developed by the Consortium will be of wide applicability in biomolecular simulation, and will be made available to the wider community e.g. through CCP-Forge. All members of the HECBioSim Management Group contribute to planning and developing the project.

Examples of HECBioSim development work and applications of these tools are listed under projects below (particularly CAL; Gareth Shannon has recently been appointed in CAL’s group at Nottingham to carry out HECBioSim development work). There are several ongoing efforts worldwide to increase scalability, throughput and productivity in biomolecular simulations. We will work with these in our developments. The tools we develop in HECBioSim will be disseminated e.g. via CCP-BioSim training workshops, our links (e.g. MW) with CCP4, CCPN, and CCP-EM, and the webpages. The aim is to develop generally useful tools for the practical integration of different types of biomolecular simulation. Work on the ProxyApp/Longbow tool, from CAL’s group, is described below.


“Core support” provided for EPSRC at Daresbury: HECBioSim receives 1 FTE of Core Support under the SLA. James Gebbie was appointed to this role in December 2013. The SLA staff effort is overseen by the Management Group. Activities and targets to date are listed below:

  • Built and launched consortium website, www.hecbiosim.ac.uk
  • Maintain information on best practice for biomolecular simulation (including code-specific guidance); introductory material for Monte Carlo codes ProtoMS (JWE) and Sire (www.siremol.org). Disseminating scripts, code and results. Develop and provide advice on optimum job parameters for biomolecular simulation software on ARCHER. Liaison with Alan Gray at EPCC on maintaining list of biomolecular software known to scale well on ARCHER.
  • Provide details of HEC allocation panel meetings and application deadlines.
  • Profile common codes/protocols on ARCHER (in terms of ns/day and disk requirements), and provide an estimate of required AUs for consortium proposals (see https://epubs.stfc.ac.uk/work/50963). Include as an option in the job submission tool.
  • Work on the job submission tool (Longbow, previously known as ProxyApp) originally developed by Charlie Laughton, which allows setup of AMBER jobs and submission to ARCHER. Introduce parameter checking to avoid common errors, or unsuitable combinations of parameters. Introduce monitoring of jobs and transfer files seamlessly between local and remote staging. Extended ProxyApp to NAMD, GROMACS, CHARMM and LAMMPS.
  • Extensions to LAMMPS code, in collaboration with Sam Genheden and Jon Essex to add the integrator from the BRAHMS code (MD program for simulation of biological membranes with the ELBA coarse-grain force field, https://code.google.com/p/brahms-md/), add improved barostat, implement multiple timestepping, and include metadynamics routines. Several UK groups are interested in LAMMPS for mixed bio/non-bio systems, but the code is currently missing some functionality available in more traditional bio codes.
  • James Gebbie has been working with PCB and Alex Heifetz (Evotec) to implement in GROMACS a new tool to assess the quality of a G-protein-coupled receptor (GPCR) model in a MD simulation. This industrial collaboration will incorporate the GLAS scoring function for GPCRs (Heifetz et al. Biochemistry 2013 52:8246), which are major drug targets. The link to the widely-used GROMACS software will ensure impact.



Example Consortium projects

we focus here on projects utilising ARCHER, including current projects; space does not permit listing all projects. See also case studies

We particularly support atomistic and coarse-grained MD simulations of ensembles, large systems and on long timescales which are only achievable through use of HEC resources. We have supported projects from across the whole UK community, including many from investigators with no prior connection with HECBioSim or CCP-BioSim, demonstrating our open approach and effective communication (see hecbiosim.ac.uk). Several projects involve industrial collaborations. We include description of some current projects below to give an indication of the breadth of work in the Consortium. These show significant progress in the relatively short amount of time since they began. Demand for ARCHER resources is growing rapidly, from existing and new users, and significantly exceeds our current allocation. We therefore request an increased allocation.

 

Sam Genheden and Jon Essex (Chemistry, Southampton) - Parameterisation of a dual-resolution coarse-grain/atomistic potential

Total AUs used to date: ~25 M. To parameterize the interaction between our coarse-grain lipid model and atomistic solutes, we have performed umbrella-sampling simulations of 17 amino acid side-chain analogues in a DOPC model membrane in the LAMMPS software. The aim is to fine-tune the interaction between the all-atom amino acids and the coarse-grained bilayer. The dual resolution model now gives results of comparable quality to atomistic studies, but at considerably reduced computational cost. We are currently converging these potentials of mean form in preparation for a publication, and are running an additional small-molecule data set by way of independent validation. The behaviour of small model peptides with the optimized dual-resolution parameter set will be explored in the next month.

 

Mario Orsi (School of Engineering and Materials Science, Queen Mary University of London) - Atomistic simulation of mixed lipid membranes

We are using the LAMMPS parallel code to simulate lipids modelled by the CHARMM force field. We are investigating fundamental properties of atomistic mixed lipid membranes, specifically, the effect of nonlamellar lipids on mechanical and electrical properties. Our results may have an impact on the current understanding of membrane proteins, and therefore on many biological processes. At this (preliminary) stage, we are planning the preparation of at least one paper based on our ARCHER calculations.

 

Jonathan Hirst (Chemistry, Nottingham) - Pore formation in a bacterial membrane by a naturally occurring antibiotic peptide, nisin

New parameters have been developed for the unusual dehydrated amino residues in nisin. Small-scale simulations of single nisin and lipid-II molecules interacting in a membrane environment have revealed a specific interaction between the N-terminus of nisin and the polar head groups of the lipid bilayer. We are now using NAMD on ARCHER to build a large scale, all-atom model of the membrane pore, consisting of 8 nisin and 4 lipid II molecules, containing ~1M atoms.

 

Marieke Schor, Giovanni Brandani and Cait MacPhee (School of Physics and Astronomy, Edinburgh) - Interfacial protein folding, 9M AUs

The bacterial hydrophobin BslA confers hydrophobicity to the B. subtilis biofilm. Our aims in this project are 1. To predict the solution structure of the cap as this is not captured in the X-ray crystal structure and 2. To characterize folding transitions of the cap in solution and at the interface. So far, we have used replica exchange solute scaling (REST2) as implemented in the GROMACS software with the PLUMED plugin to predict the cap structure in solution, giving insight into the stability of the folded and unfolded cap conformations in solution and at the interface.

 

Julien Michel (Chemistry, Edinburgh) - Elucidation of ligand dependent modulation of MDM2 lid dynamics

MDM2 is a protein implicated in the progression of many forms of cancer and a potential target for new anti-cancer agents. Only the structure of the MDM2 ‘’core’’ region is known from experiments. We are using large-scale accelerated MD and umbrella sampling MD simulations with AMBER to determine the structure of a missing element, the lid region. The simulations show that the lid region is very dynamic, which explains why it is difficult to determine the structure of this region with experiments. The simulations also show that the lid dynamics is strongly influenced by the presence of different small-molecule ligands and have explained why the piperidone ligands order the MDM2 lid upon complex formation. The results of the study are of direct relevance to preclinical drug discovery efforts for this important cancer target. The work is under revision at PLOS Computational Biology.

 

Francesco Gervasio (UCL) - Studying the Effects of Oncogenic and Drug-Resistance-inducing Mutations on the free energy landscape of Protein Kinases

We have seen significant changes in the activation dynamics and free energy landscape due to mutations and have characterized the drug-binding mechanisms, using GROMACS 4.6 and 5.0, PLUMED 1.3 and 2.1. The results have been presented at several national and international meetings (ACS annual meetings, San Francisco Aug 2014, Denver March 2015; Biophysical Society, Istanbul, Sept. 2014. A manuscript has been submitted to JACS and another to PNAS. Collaborations: UCB Pharma and Cancer UK Manchester (with Prof. R. Marais, director of the institute). Impact: design of new drugs less prone to of drug-resistance in melanoma (B-Raf) and leukemia (Src).

 

Richard Sessions and Dek Woolfson (Biochemistry/Chemistry, Bristol).

The 23,000 kAU allocation of time from the HECBioSim consortium has enabled us to run out the atomistic simulation of a designed SAGE particle (Fletcher et al. Science 340 595, 2013) encapsulating ATP molecules to from 20 to 50 ns, leading to further escape of the nucleotide. Similarly, we have extended the “empty” SAGE simulation a further 50 to 65 ns. Extending these simulations is vital to understanding of the conformational behaviour of the hetero-dimer peptides that couple the trimer-peptide hubs. Rotation of the dimer around the linking disulphide bonds influences the curvature expressed by the hexagonal net. In turn, this is one of the factors governing SAGE particle size. These SAGE particles contain about 93000 amino acids and the simulation comprises 44 million atoms (see figure); these very large atomistic MD simulations that are only possible on a machine like ARCHER. We are also exploring the effect of sequence alterations in the dimer peptide upon curvature by simulation of patches comprised of 19-hexagon nets. The Woolfson group is continuing to explore the experimental behaviour of such constructs in tandem with our calculations.

 

Syma Khalid (Chemistry, Southampton) - Antimicrobial peptides are cationic proteins that can induce lysis of bacterial cells through interaction with their membranes.

Current models for the mechanisms of cell lysis tend to neglect the role of the chemical composition of the membrane, which differs between bacterial species and can be heterogeneous even within a single cell. We are performing, to our knowledge, the first atomistic MD simulation of the interaction of an antimicrobial peptide, polymyxin B1 with chemically complex models of both the inner and outer membranes of E.coli. The results of >16 microseconds of simulation predict that polymyxin B1 interacts with the membranes via distinct mechanisms. The simulations we have performed on ARCHER are providing key insights into the mechanisms of action of antibacterial agents. This work is currently under review at PLoS Comp. Biol.

 

Edina Rosta (Chemistry, KCL) - Role of dimerization in the activation mechanism of RAF kinases.

Single-residue mutations of BRAF are linked to 66% of malignant melanomas, and also to ovarian, colon and papillary thyroid cancers. We have carried out MD simulations of phosphorylated BRAF and CRAF homo- and heterodimers, revealing major structural rearrangements in the protomers. Based on our findings, we established an collaboration with the group of Prof. Walter Kolch, Director of Systems Biology Ireland and Conway Institute. They carried out mutational experiments to validate key aspects of the conformational transitions and MD structures. Our results shed new light on the ubiquitous kinase activation mechanisms involving homo- and heterodimers. We are preparing our manuscript for. Our results have been presented at seminars in the UK and overseas (Oxford; NIH(US) NHLBI and NIDDK institutes; D. E. Shaw Research in New York; ACS National Meeting, Denver, March).

 

Phil Biggin (Biochemistry Oxford) - 1. Understanding the Gating of GABAA Receptors. 2. Accurate calculations of absolute binding free energies and enthalpic/entropic contributions for anticancer compounds.

The first project involved long (several hundred nanoseconds) MD simulations on a large neurotransmitter receptor. Understanding how these receptors move between their different (gated) states is of major interest, especially for improved treatments for diseases like epilepsy. The second project has allowed us to perform accurate (less than 2 kcal/mol MSE) free energy calculations. Because of the extensive sampling required to achieve convergence via the use of Hamiltonian Replica Exchange calculations, a large number of cores were required. Our results suggest that absolute free energy calculations on clinically approved drugs are accurate enough to be useful in a lead-optimization program. The results of this work will presented in two talks at the Biophysical Society Annual Meeting in Baltimore, Feb. 2015 and in posters at the the Annual CCP-BioSim Conference, Leeds, January 2015.

 

Sarah A. Harris, and Agnes Noy (Physics, Leeds) - Comparison of different DNA molecular contours in measuring the writhe of supercoiled DNA

Evaluation of writhe in supercoiled DNA using MD simulations of minicircles with 260 and 336 base-pairs. We provide a measure of DNA writhe suitable for comparing atomistic resolution data with those obtained from the global molecular shape. The local 3DNA framework largely overestimates writhe due to the basal local periodicity of DNA. However, this local writhe seems to emerge as an internal mechanism for the DNA molecule to confront superhelical stress, that oscillates between low and high twist structures, coupled to a high and low local periodicity, respectively, mimicking the different conformational space of A and B canonical DNA forms.

 

Valeria Losasso, Hannes Loeffler, Martyn Winn (STFC Daresbury Laboratory) - Simulations of the intra-cellular domains of EGFR

Requested AUs: 3.9M; Code used: NAMD 2.9. As part of a BBSRC-funded experimental/computational consortium on cellular signalling networks (BB/G006911), we have been performing MD simulations on Epidermal Growth Factor Receptor (EGFR), involved in the regulation of cell proliferation, differentiation, migration, and apoptosis. Misregulation of EGFR signalling pathways have been implicated in several cancer types. We have focused on linkage between the trans-membrane helices and the intra-cellular juxta-membrane domains, and their relation to the plasma membrane. 200ns-500ns MD simulations of 8 systems show how palmitoylation leads to destabilisation of the inactive conformation. These observations are being related to experimental measurements of EGFR conformations (via FRET) and activity, to understand the regulatory role of the juxta-membrane domains.

 

Andrei Pisliakov (Dundee) - MD Simulations of Respiratory Complex I

In this project we aim to identify the proton transfer pathways in Complex I, a respiratory enzyme and proton pump that couples quinone reduction to proton translocation across the membrane, and thus initiates the processes of energy production in mitochondrial and bacterial respiratory systems. We have carried out all-atom MD simulations of Complex I (simulation system size ~400,000 atoms) using NAMD for a combined simulation time of nearly 1.5 microseconds. We have characterized four plausible proton transfer pathways in the membrane domain of Complex I and proposed the molecular mechanisms of the proton uptake/release to the cytoplasmic and periplasmic sides of the membrane. Collaboration with Leo Sazanov, MRC Mitochondrial Biology Unit, Cambridge; our results provided guidance for new experiments.

 

Jiayun Pang (Greenwich) - Dynamical Control of Radical Intermediates in the Vitamin B12-Dependent Enzymes

This project focuses on understanding the catalytic strategies employed by the family of AdoCbl-dependent enzymes to reinforce the control of highly reactive radical intermediates. Codes used: CPMD/GROMOS QM/MM code.

 

Dr Tomasz Włodarski, Prof. John Christodoulou (Institute of Structural Molecular Biology, UCL and Birkbeck College) - Computational study of co-translational protein folding and misfolding on the ribosome

Aberrant protein folding often results in misfolding and aggregation, and can lead to a variety of human disorders, including Alzheimer’s and Parkinson’s diseases. By combining MD simulations with NMR structural restraints we are producing atomic structures of ribosome nascent chain complexes. We are aiming to investigate single and multi-domain protein folding on the ribosome. We conduct all-atom MD simulations of different stages of co-translational folding of ddFLN5 for which we have obtained an extensive set of NMR data, representing different stages of the biosynthesis and the folding of ddFLN5. Simulations are carried in Gromacs 5.0.4 patched with PLUMED 2.1 and Almost, with the Amber03w forcefield and combined bias-exchange metadynamics, enabling more efficient sampling of the free energy landscape, with the use of NMR data (chemical shifts and RDCs) as structural restraints. This project will provide unprecedented atomistic details of the process of co-translational protein folding.

 

Michele Vendruscolo (Cambridge Chemistry) - Conformational sampling of α-synuclein by NMR-restrained bias exchange metadynamics simulations.

19,600 kAUs. The is to obtain an accurate picture of the conformational space of α-synuclein, a 140-residue intrinsically disordered protein whose aggregation process is related with Parkinson’s disease, applying the latest development of the replica-averaged metadynamics (RAM) scheme, recently implemented in our group. Based on MD enhanced sampling techniques (well-tempered bias-exchange metadynamics), the approach includes NMR data using the maximum entropy principle to improve the accuracy of the simulation. We use Gromacs 4.6.7, with the open source library PLUMED 2.1 and the ALMOST code. The simulation (now in full production phase) is currently running on 64 ARCHER nodes (1536 cores). We have used about 8,500 kAUs so far and will use the entire allocation by the end of March.

 

Charlie Laughton et al., Pharmacy, Nottingham - Ensemble simulations of protein dynamics and protein-ligand recognition

One of the biggest challenges in biomolecular simulation is to adequately sample the conformational space such that statistically converged metrics – such as free energies of binding – can be calculated from the data. Traditionally, the approach this was to run a single simulation for as long as possible, but it is now clear that for many purposes it is just as effective, and computationally more tractable, to run many copies of the same simulation concurrently, with randomised initial conditions such that they explore conformational space independently. Tools to simplify the setup and analysis of such simulations – no trivial issue as this quickly becomes a “big data” problem – are under development as part of the HECBioSim project.

Here we outline a number of related projects making use of this methodology:

1. Ligand-induced changes in protein flexibility. Extending previous work (Roy & Laughton, Biophys J 2012), we have performed 100 independent 100ns simulations on the mouse major urinary protein (MUP) in the presence and absence of a pyrazine ligand. This work was presented at the CECAM workshop on Entropy in Biomolecular Systems (Vienna, May 2014) (paper under review in Bioinformatics). Codes used: AMBER, GROMACS.
2. MM-GBSA calculations of protein-peptide binding affinities. As part of a Kurdish-government sponsored PhD project on developing inhibitors of the TRF1-TIN2 interaction in telomeres, we are using the MM-GBSA method to predict protein-peptide binding affinities. We find that the ensemble simulation strategy gives significantly improved predictions of relative binding affinity, confirmed by synthesis and biophysical evaluation of selected TIN2-derived peptides. Codes used: AMBER, GROMACS.
3. Exploring allostery in beta-secretase (BACE) inhibition. In collaboration with Professor Javier Luque, University of Barcelona, we have used ensemble simulation approaches to analyse the conformational flexibility of BACE-1, revealing the spontaneous opening of an allosteric pocket that supports the mode of action of a new class of enzyme inhibitors. The newly identified conformations of the protein are now being used as targets for docking studies, and selected molecules are being synthesised ready for biological evaluation.
4. Exploring IgE with MSM ensemble simulations (Industrial Collaboration with Ben Cossins, UCB). UCB are interested in large simulations to understand the dynamics of protein drug targets. In this study, the Markov state model (MSM) approach along with time independent component analysis (TICA) is used to analyse ensembles of simulations of Immunoglobulin E (IgE). Owing to the size and dynamic nature of this protein, it is very hard to probe with methods such as NMR, therefore this is a good example of the value MD can provide.

 

Mark Sansom (Oxford) - Modelling protein crowding in bacterial membrane

In Gram-negative bacteria, embedded proteins, called porins, fulfil crucial tasks such as letting proteins and small molecules translocate through the membrane. In collaboration with the experimental group of Colin Kleanthous, we are studying the complex dynamic behaviour of large patches of E. coli membrane. We have performed coarse grained simulations using the MARTINI in GROMACS of two different porins, BtuB and OmpF, present in high concentrations at the surface of E. coli. These systems will push the boundaries of the modelling to its very limits using currently available MD programs. To evaluate the feasibility of such large systems, we have created membrane models of 120 nm and 240 nm side, 4 times the size of models previously studied.

 

Management

The HECBioSim Management Group is listed above, and meets regularly. HECBioSim also takes advantage of the strong management and advisory structures of CCP-BioSim. The Advisory Board was expanded and refreshed for the renewal and consists of: Dr. Nicolas Foloppe (Vernalis plc, Chair); Dr. Colin Edge (GlaxoSmithKline); Dr. Mike King (UCB Pharmaceuticals); Dr. Mike Mazanetz (Evotec AG); Dr. Garrett Morris (Crysalin Ltd.); Dr. Gary Tresadern (Johnson and Johnson Pharmaceuticals); Dr. Richard Ward (AstraZeneca); Prof. Modesto Orozco (IRB, Barcelona); Prof. Tony Watts, (Oxford, NMR); Dr. Pete Bond (A*STAR Bioinformatics Institute Singapore).

Details of application procedures are provided through our website, hecbiosim.ac.uk. A resource allocation panel, with rotating membership, decides on allocation of resources to members and non-members of the Consortium. Decisions are made by consensus wherever possible, with majority voting (with the Chair having a casting vote) when required. The resource allocation panel meets 3 times per year, to allocate a third of the annual compute resource for full applications. Membership of the Panel rotates regularly, with two new members joining for each meeting, with the Chair of the next Panel chosen from one of the previous non-chairing members for continuity. The proposals are ranked according to scores awarded by the Panel, with primary criteria of scientific excellence, technical feasibility and track record, suitability for ARCHER, community building and impact, including industrial impact. We encourage non-traditional users of HEC and early career researchers via ‘light touch’ refereeing of small ‘pump-priming’ applications (1M AUs for 3 months) by one Panel member; the expectation will be that small proposals will be supported, based on technical feasibility and scientific interest. Advice on testing, scalability and development will be provided via the SLA Support, helping to develop small proposals into full proposals.

 

Measuring Quality of Output; Review of Progress and Scientific Priorities

Scientific and computational priorities are set by the Management Group. Assessment and monitoring of progress and quality of outputs is carried out also by the Advisory Board (see above). Use of ARCHER time is monitored in contact with national HEC Centres. A condition of each award is that the applicant is required to submit a short report on usage and outputs. Acknowledgement of HECBioSim and EPSRC in resulting publications and presentations is a condition of the award of ARCHER time. Projects are invited to present the scientific outputs of their work at CCP-BioSim conferences (ccpbiosim.ac.uk), aiding dissemination.

 

Future plans

We are oversubscribed and so request an increased allocation. We aim to expand the breadth of the work of the Consortium focusing on cutting-edge applications, and building collaborations with experiments and industry, to achieve maximum impact from ARCHER use.