Annual Report for EPSRC High End Computing Consortia 2016/17

HEC Consortium

HECBioSim (The UK High-End Computing Consortium for Biomolecular Simulation)

Consortia Chair

Adrian Mulholland (School of Chemistry, University of Bristol)

Management Group

Prof Syma Khalid (University of Southampton) (Chair)
Dr Julien Michel (University of Edinburgh) (secretary)
Prof Adrian Mulholland (University of Bristol)
Prof Philip Biggin (University of Oxford)
Prof Jonathan Essex (University of Southampton)
Prof Francesco Gervasio (University College London)
Dr Sarah Harris (University of Leeds)
Dr Richard Henchman (University of Manchester)
Prof Charles Laughton (University of Nottingham)
Dr Edina Rosta (King’s College London)
Prof Mark Sansom (University of Oxford)
Dr Irina Tikhinova (Queens University, Belfast) 

Background

Molecular simulations on high performance computing resources can show how biological molecules 'work' by providing details at the level of individual atoms. For example, molecular simulations can show how drugs bind to their biological targets, how enzymes catalyse reactions, and how proteins, nucleic acids and other molecules interact with each to maintain healthy cells. Biomolecular simulation is a vibrant and growing area of international importance, making increasingly significant contributions to biology, biotechnology and healthcare. HECBioSim brings HEC to a wider UK community which include non-traditional users and experimental bioscientists. In addition, we engage and work with physical and computer scientists in bringing their methods to biological applications. We are an open consortium, any UK PI working in the area of biological simulations can apply to us for time through our rigorous panel-reviewed evaluation system. Work resulting from ARCHER time allocated through HECBioSim has been published in Nature, Science and Nature Structural and Molecular Biology. HECBioSim began in 2013 under the leadership of Adrian Mulholland and was renewed in 2018 with Syma Khalid as chair. From 2018, HECBioSim is also responsible for allocating time for biomolecular simulations on the Tier 2 resource, JADE and more recently we have also started testing and bechmarking Bede.

Workshops and New Opportunities

We are currently in the process of planning a series of workshop for beginner users of HPC resources to introduce them to the HPC resources available inthe UK at Tier1 and Tier2 levels. These will focus on teaching about the differences in architectures, what they are best for and how to get the most performance from their simulations. It is anticipated these will form part of next years delivery programme.

We are also in the process of planning a set of workshops on the subject of how to install your own software on the various HPC platforms in the UK and how to optimise them for performance.

Issues and Problems

None

Membership

Please provide a full list of existing members and their institutions, highlighting any new members that have joined the consortium during the reporting period. If available please provide information on the number of distinct users that have accessed ARCHER via the Consortium during this reporting period.

Below is a list of PIs with HECBioSim projects in the current reporting period:

Prof Philip Biggin (University of Oxford)
Prof Jonathan Essex (University of Southampton)
Dr Phillip Fowler (University of Oxford)
Prof Francesco Gervasio (University College London)
Dr Daniel Kattnig (University of Exeter)
Prof Syma Khalid (University of Southampton)
Dr Alessandro Pandini (Brunel University, London)
Prof Maria Carmen Romano (University of Aberdeen)
Prof Mark Sansom (University of Oxford)
Prof Edina Rosta (UCL)P
rof Adrian Mulholland(Bristol)
Prof Steve Matthews (Imperial College London)
Prof Charlie Laughton(University of Nottingham)
Prof D. Flemming Hansen (UCL) –new during this reporting period.
Dr Gabriele Sosso (University of Warwick)
Dr Phillip Stansfeld (University of Warwick)
Dr Antreas Kalli (University of Leeds)
Dr Arianna Fornili (Queen Mary University London)
Dr Edina Rosta (King’s College London)
Prof Franca Fraternali (King’s College London)
Prof John Christodoulou (University College London)
Dr Martin Ulmschneider (King’s College London)
Dr Michelle Sahai (University of Roehampton London)
Dr Marc van der Kamp (University of Bristol)
Dr Agnes Noy (University of York)
Dr Sarath Dantu (Brunel) –new during this reporting period (ECR)
Dr Valeria Lossaso (STFC)-new during this reporting period. 

During the November 2019 to October 2020 reporting period, a number of high-profile publications have arisen from work conducted by the Consortium, including in Science, Science Advances, Natureand Nature Comms. Our science has resulted in numerous invited international talks and a number of successful grant applications. Our work on covid has made national newspapers and has resulted in a prestigious international award, details of both are provided within this annual report.

All papers are in researchfish.

A few highlights include:

Yamamoto E, Domański J, Naughton FB, Best RB, Kalli AC, Stansfeld PJ, Sansom MSP.(2020) Multiple lipid binding sites determine the affinity of PH domains for phosphoinositide-containing membranes, Science Advances

Joanna Szczepaniak, Peter Holmes, Karthik Rajasekar, Renata Kaminska, Firdaus Samsudin, Patrick George Inns, Patrice Rassam, Syma Khalid, Sean M. Murray, Christina Redfield& Colin Kleanthous (2020) The lipoprotein Pal stabilises the bacterial outer membrane during constriction by a mobilisation-and-capture mechanism, Nature Comms.

The emergence of sequence-dependent structural motifs in stretched, torsionallyconstrained DNA Jack W Shepherd, Robert J Greenall, Matt I J Probert, Agnes Noy, Mark C Leake. Nucleic Acids Research, Volume 48, Issue 4, 28 February 2020, Pages 1748–1763,

V Velasco-Berrelleza, M Burmann, JW Shepherd, MC Leake, R Golestanian, A Noy(2020). “SerraNA: a program to determine nucleic acids elasticity from simulation data” Phys Chem Chem Phys, 22, 19254-19266 https://doi.org/10.1039/D0CP02713H This is the citation for new software from Noy and Harris to calculate global elastic constants from DNA fragments using all-atom MD. The related software is at: https://github.com/agnesnoy/SerraNA

NJ Cook, W Li, D Berta, M Badaoui, A Ballandras-Colas, A Nans, A Kotecha, E Rosta, AN Engelman, P Cherepanov, “Structural basis of second-generation HIV integrase inhibitor action and viral resistance”, Science, 10.1126/science.aay4919, 2020

The HECBioSim key code is Longbow(downloaded 8,633 times). Longbow functions to seamlessly offload simulations to powerful super computers in a manner that appears to the user as the simulation had run on their own PC. In our community, users fit broadly into two groups as far as HPC is concerned. The expert users and those that have never used any big machines, and there is nothing in between.

For the experts, longbow enables them to rapidly submit large volumes of simulations without spending long tedious hours making submit scripts and transferring files. This can all be done automatically with a single command. For the beginner users, and these include non-linux folks and experimentalists dabbling with simulation. These folks do not understand the commandline tools or which ones to use for what, however they do know how to run basic md simulations on their own computer. Longbow lowers the technical barriers to using larger HPC machines by abstracting this all away.

Development in this reporting cycle has focused on working towards a new major release that adds support for ARCHER2 and the second generation of Tier2 machines along with the existing onesthatare continuing their service. A development priority has also been an enhanced security model to be compatible with the new hardened security adopted by ARCHER/ARCHER2 and a number of Tier2s. Other development has seen the addition of support for more scientific codes, namely Chemshell andDL_Poly, the addition of support for the latest python versions 3.7 and 3.8. Support added for running Longbow in Docker based environments and fixes for a number of userreported errors and better documentation has been written.

Another initiative we are running is a UK wide look at the available HPC resources.We aniticipate this initiative will lead to greater productivity and more results for a given compute allocation.This entails an enhanced benchmarking and optimisation run on each of the machines we have access to where we are running various optimisation strategies across all our consortium supported codes (AMBER, GROMACS, LAMMPS and NAMD)to find the best performance configurations for each architecture. We currently have done this on ARCHER, ARCHER2, JADE, JADE2, Bede, Thomas, ISAMBARD, Dirac. We are currently working on sharing this information via a new intuitive and visually aesthetic section on our website that will provide benchmark figures and example job submission scripts along with information on how the performance levels were produced.

The number of of PhDs and Post-Docs that have been trained in the use of ARCHER during the reporting period exceeeds 20.

Due to the pandemic that has affected nearly all of the planned training events in this reporting period. The consortium has focused on delivering training online using our new state of the art cloud based training platform.

There is a training module that is freely available online focussed on training users in the use of Longbow with ARCHER. This training module has been accessed 45 times during this reporting period.

HECBioSim contributed a training day as part of the CCPBioSim training week in September 2020 entitled “An Introduction to Software and Hardware for Biomolecular Simulation”which was introduction to the kinds of software and hardware available for biomolecular simulation and how to use them. This course was aimed at beginners that had just started PhDs and was attended by 187 individuals from around the world.

HECBioSim Chair Syma Khalid gave a talk at the CCP4 training weekend in which she introduced HPC to an experimental (structural biology) audience.

Examples of industrial engagement through HECBioSim include:

Sansom (Oxford):

UCB have supported a number of BBSRC iCASE studentships in the general area of membrane protein/lipid interactions, all of whichhave benefitted from ARCHER access. These molecular interactions play a key role in modulating the function of membrane proteins implicated in many diseases, and therefore are of importance to pharmaceutical companies such as UCB.

Sansom (Oxford) and Stansfeld (Warwick):

IBM support an EPSRC iCASE studentship to examine peptide /lipid interactions which also benefits from ARCHER access.This has resulted in a preprint within the funding period: (which is now published, but just after this reporting period).

Biggin (Oxford)We have been extending our work on absolute binding free energies to fragment design with Boehringer Ingelheim, via a fully-funded PDRA. New code has been developed as part of this collaboration, which will be made available soon. AniCASE studentship on membrane proteins is funded by Vertex. Both projects have been built on the successful usage of ARCHER via HECBioSim.

Khalid (Southampton)

A studentship funded through Oxford Nanopore Technologies Ltd relies upon access to ARCHER via HECBioSim. We are simulating the protein used in the commercial devices made by the world leading nanopore DNA sequencing company. Much of the work cannot be published due to the highly sensitive nature of the details of the engineered protein being used. However a paper on a model system is on biorxiv (now published, but outside of this reporting period) and a second on the wildtype protein complex CsgG/F is in preparation with the Steve Matthews and Sarah Rouse groups at Imperial College (also HECBioSim members).

Essex (Southampton)

A PhD studenthas been exploring the conformations of the recognition loops of antibodies to understand how they achieve such high affinity and specificity. This work is supported by UCB

Another PhD student is using archer to simulate the protein complex apoferritin in very large volumes of water, to provide model data to explore the accuracy of cryo-EM image reconstruction algorithms. This is in collaboration with the Rosalind Franklin Institute.Publications from both of these projects are in the pipeline.

Michel (Edinburgh)

PhD student funded by Cresset.The Michel group and Cresset are using HPC provided through HECBioSim to develop machine-learned (ML) models of the difficulty of protein ligand binding Free Energy Perturbation (FEP) calculations. The ML models are used to power new automated FEP workflows for high-throughput studies that support industrial drug design R&Defforts.

Fraternali (King’s College London)

A novel antimicrobial nanocapsule construct able of self assembly. The results are presented in the manuscript under revision “Nanocapsule Designs for Antimicrobial Resistance”. The project carried on previous work performed incollaboration with the National Physics Laboratory(UK), and prompted a collaboration with Unilever (UK) for the investigation of analogous antimicrobial peptides interactions and self-assembly. Unilever provided funding for a PDRA, access to ARCHER via HECBioSim played a role in securing the funding and setting up the collaboration. Such molecules are of interest for the company, aiming at incorporating antimicrobial elements into its products. Two manuscripts are resulting from this collaboration, one is in second revision and the other is in preparation.

Stansfeld (Warwick)

PhD studentship funded by OMASS has yielded the paper: Ligand induced conformational dynamics of the LPS translocon LptDE. Francesco Fiorentino, Joshua B. Sauer, Xing Yu Qiu, Robin A. Corey, C. Keith Cassidy, Benjamin Mynors-Wallis, Shahid Mehmood, Jani Reddy Bolla, Phillip J.Stansfeld, Carol V. Robinson (2020) Nature Chemical Biology. The simulations reported here were performed on ARCHER with access provided by HECBioSim.

Gervasio (UCL)

Collaboration with Ben Cossins (UCB, Slough) on developing accurate algorithms for ligand binding free energy calculations.

Collaboration with Astra Zeneca (who are co-sponsoring an industrial student with EPSRC): understating the effects of cancer-causing mutations on the structure and dynamics of EGFR.

Examples of HECBioSim international engagement include:

Collaboration of Khalid with Rommie Amaro (UCSD, USA) and Michael Feig(Michigan, USA)

New collaboration initiated in April 2020 to model the M protein of the SARS-Cov2 virus.The structure of this protein (the most abundant protein in the viral membrane) is not known yet it is essential for the structural integrity of the viral particle. Khalid, Amaro and Feig are using bioinformatics, molecular modelling and multiscale simulations to predict the structure of the protein and its arrangement within the viral membrane including around the spike protein. The coarse-grained simulations are being run on Archer while the atomistic simulations are being run on Frontera at the Texas Advanced Supercomputer Center. A manuscript is in preparation. The collaboration of Khalid with Amaro’s team and other US teams resulted in the award of the Gordon-Bell Special Prize for Covid19 related work. This made mainstream news in the USA.A preprint is available here: doi: https://doi.org/10.1101/2020.11.19.390187

Collaboration of Khalid with Wonpil Im(Lehigh, USA)

Khalid has been collaborating with Wonpil Im(Lehigh, USA) to use the hydrogen mass repartitioning technique to enable faster simulations of bacterial membranes. The preprint was uploaded during the reporting period but the paper came out just after so will be included in the next annual report. This work represents an important advancement as it enables users on ARCHER to use longer time steps with the CHARMM force-field thereby speeding up simulations.We expect this study to be highly cited given its methodological utility and that the time-step can be set-up via the free online server CHARMMGUI.

Collaboration of Khalid with Gizela Storz (NIH, USA)

This is a collaboration that is also involves Ben Luisi (Cambridge) and brings together multiple experimental methods with computer simulations to understand protein-protein and protein-lipid interactions in a bacterial membrane environment.

Du D, Neuberger A, Orr MW, Newman CE, Hsu PC, Samsudin F, Szewczak-Harris A, Ramos LM, Debela M, Khalid S, Storz G, Luisi BF (2020), Interactions of a Bacterial RND Transporter with a Transmembrane Small Protein in a Lipid Environment. Structure

Collaboration of Sansom withSudha Chakrapani, (Case Western U, Cleveland,USA)

ARCHER allocations through HECBioSimwere used for a number of joint studies of ion channel structure function relationships with both collaborators leading to the following publications:

Kumar, A., Basak, S., Rao, S., Gicheru, Y., Mayer, M.L., Sansom, M.S.P. & Chakrapani, S. (2020) Mechanisms of activation and desensitization of full-length glycine receptor in membranes.Nature Communicatons.

Collaboration of Sansom, Kalli and Stansfeld with Eiji Yamamoto (Keio University, Japan) and Robert Best(NIH, USA).

Further details are in one of our case studies.

Yamamoto E, Domański J, Naughton FB, Best RB, Kalli AC, Stansfeld PJ, Sansom MSP.(2020) Multiple lipid binding sites determine the affinity of PH domains for phosphoinositide-containing membranes, Science Advances

Collaboration of Sansom with David Baker(University of Washington, USA)

This is a newly established collaboration in which the combined expertise of the UK and USA groups are being used for de novo design of membrane protein pores. A PhD student (Shanlin Rao) from the Sansom group is being trained as part of the project, which from the UK side relies on HPC provided via HECBioSim.

Stansfeld - multiple international collaborations:

Jochen Zimmer (Virginia, US)
Linda Columbus(Virginia, US)
Inga Hanelt (Frankfurt, Germany)

Structural basis of the proton-coupled potassium transport in the KUP family Igor Tascón, Joana Sousa, Robin Corey, Deryck Mills, David Griwatz, Nadine Aumüller, Vedrana Mikusevic, Phillip J. Stansfeld, Janet Vonck, Inga Hanelt,Nature Communications

Cristina Paulino (Groningen, Netherlands)
John Parkinson (Utah, US)
Irina Gutsche (Grenoble, France)

Complete structure of the core signalling unit of the E. coli chemosensory array in an optimised minicell strain. Alister Burt, CK Cassidy, Peter Ames, Maria Bacia-Verloop, Megghane Baulard, Karine Huard, Zaida Luthey-Schulten, Ambroise Desfosses, Phillip J. Stansfeld, William Margolin, John Parkinson, Irina Gutsche. Nature Communications,2020

Filippo Mancia (Columbia, US)
Jonathan Dworkin (Columbia, US)
Ed Yu (Case Western, US)
Multiple international authors from UK, Slovenia, USA
Structure and mechanism of bactericidal mammalian perforin-2, an ancient agent of innate immunity, Tao Ni, Fang Jiao, Xiulian Yu, Saša Aden, Lucy Ginger, Sophie Williams, Fangfang Bai, Vojtěch Pražák, Dimple Karia, Phillip Stansfeld, Peijun Zhang, George Munson, Gregor Anderluh, Simon Scheuring, Robert J. C. Gilbert, Science Advances, 2020

Gervasio withKarl-Heinz Krause (Geneva):

Collaboration with Prof. Karl-Heinz Krause, faculty of Medicine University of Geneva. Understanding the effect of Mutations in SARS-CoV2 Spike protein. Preprint available at https://www.medrxiv.org/content/10.1101/2020.08.23.20180281v1, submitted for publication).

Mulholland with Amaro(UCSD)

Biomolecular simulations for covid applications, highlighting HECBioSim activity, are reviewed in: ‘Biomolecular Simulations in the Time of COVID-19, and After’ R.E. Amaro and A.J. Mulholland Computing in Science & Engineering Nov.-Dec. 2020, pp. 30-36, vol. 22 DOI: 10.1109/MCSE.2020.3024155

COVID19 -Computational Chemists Meet the Moment, Adrian J. Mulholland and Rommie E. AmaroJ. Chem. Inf. Model. 2020.

Simulations support the interaction of the SARS-CoV-2 spike protein with nicotinic acetylcholine receptors A. SofiaF. Oliveira,Amaurys AvilaIbarra,IsabelBermudez,LorenzoCasalino,ZiedGaieb,Deborah K.Shoemark,TimothyGallagher,Richard B.Sessions,Rommie E. Amaro,Adrian J.Mulholland https://www.biorxiv.org/content/10.1101/2020.07.16.206680v3

Mulholland with multiple international authors:

Free fatty acid binding pocket in the locked structure of SARS-CoV-2 sspikeproteinScience 370 725-730(2020) https://science.sciencemag.org/content/370/6517/725

We would like to highlight here theworld-leading work of the UK biomeolcular simulation community on covid19.

The HECBioSim management group played a big role (at short notice and often during the summer break) in helping EPSRC to evaluate HPC proposals on covid19 work, thus helping to prioritise the most promising for support on ARCHER and andthe tier two resource JADE. With additional high quality proposals coming in a bit later we were able to liaise with the Cambridge HPC team and the N8 team for access to their resources (Vendruscolo, Cambridge, will give a short presentation on his Covid 19 work at the Bede launch event on January 28th).

We set up a website to bring together UK based researchers with overseas researchers and a directory of available resources and news stories to maximise research efficiency and complementarity of expertise (further described in one of our case studies). In the early days of lockdown and research into covid19 this website proved an invaluable resource.

The PIs we have supported on Covid projects are:

Essex (Southampton)
Biggin (Oxford)
Sansom (Oxford)
Duarte (Oxford)
Michel (Edinburgh)
Rosta (KCL, now UCL)
Vendruscolo (Cambridge)
Gervasio (UCL)
Haider (UCL)
Fraternali (KCL)
Ulmschneider (KCL)

Additional time has been provided where needed leading to some large, and important collaborations, for exampleVendruscolo is working with the Rosalind Franklin Institute to study the dynamics of antibodies that bind the viral spike protein(a follow on of the project described here: https://www.rfi.ac.uk/engineered-llama-antibodies-neutralise-covid-19-virus/). Both Vendruscolo and Mulholland are currently using time on Bede and ARCHER2 respectively allocated via evaluation by HECBioSim to study the newly emerged mutations in the spike protein (UK and South Africa strains).New industrial and overseascollaborations have been established in which simulations performed on ARCHER are facilitating the development of anti-virals that target the SARS-CoV2 virus. These include:

Gervasio (UCL): collaboration with Mone Zaidi, Mount Sinai School of MedicineandDeepMind -London (Designing therapeutic peptides for COVID).

Biggin (Oxford): collaboration with IBM on designing antivirals targeting SARS-CoV2 enzymes.

Mulholland (Bristol) collaboration with: Geneva Biotech Sàrl and Imophoron Ltd to characterise druggable sites on the SARS-CoV2 Spike protein.

The UK projects on covid19 work have increased the worldwide reputation and visibility of the UK’s research in biomolecular simulations. This is also helped by authoratative topical reviews co-authored by Mulholland:

Biomolecular simulations for covid applications, highlighting HECBioSim activity, are reviewed in: ‘Biomolecular Simulations in the Time of COVID-19, and After’ R.E. Amaro and A.J. Mulholland Computing in Science & Engineering Nov.-Dec. 2020, pp. 30-36, vol. 22 DOI: 10.1109/MCSE.2020.3024155

COVID19 -Computational Chemists Meet the Moment, Adrian J. Mulholland and Rommie E. AmaroJ. Chem. Inf. Model. 2020.

Khalid’s collaboration with Amaro in which the know-how gained from running simulations on ARCHER was used to advise the Amaro group and helped to secure the Gordon-Bell Special Prize for Covid19 work: https://www.hpcwire.com/2020/11/19/gordon-bell-special-prize-goes-to-massive-sars-cov-2-simulation/

Overviews of the work of the consortium in this area have been presented as videos by Khalid, see: https://epsrc.ukri.org/newsevents/multimedia/how-supercomputers-are-helping-in-the-fight-against-covid-19/and:https://www.archer2.ac.uk/training/courses/201118-covid19-webinar/

D. Flemming Hansen (UCL) is a PI who is new to HPC, he was able to access ARCHER and support in terms of running simulations on HPC resourecs via HECBioSim and has now gone onto secure a Pioneer grant for ARCHER2 time. He says:

The HECBioSim allocation for our project “Drug Discovery for Viral Disordered Proteins” allowed us to perform enhanced sampling molecular dynamics simulations of a disordered protein from the hepatitis C virus using data from nuclear magnetic resonance spectroscopy as restraints to improve accuracy. We are currently finishing these simulations on our local cluster at UCL before submitting the results for publication. The simulations performed on ARCHER provided us with essential results that has allowed us to secure a two-year PIONEER grant on the ARCHER2 supercomputer, which in turn will allow us to scale up our work.

Gabriele Sosso(Warwick):

An early career researcher at Warwick was provided initially with pump-priming time on ARCHER and then helped to put together a successful full application for the project: “The atomistic details of the Ice Recrystallisation Inhibition activity of PVA”.He says: “Access to ARCHER (via the HECBioSim grant) has been key to deliver this workas we needed an awful lot of time to converge our statistics. In fact, to my knowledge this is the first time that someone has made the effort to study theinteraction between polymers anddynamically growingice surfaces (takinginto account different surface coverages as well, which means alotof computer time)”. The resulting work has now been accepted in Nature Comms, but is not yet published (will be in the annual report for next year).Furthermore the data has helped Sosso put together a UKRI FLF application with HECBioSim chair, Khalid as his external mentor.

Antreas Kalli (Leeds):

An early career researcher at Leeds has used ARCHER time to:

“Enhance our collaboration with Professor David Beech (Univeristy of Leeds) to study the Piezo1 mechanosensitive channel at the molecular level. Piezo1 is a recently discovered mechanosensitive channel that senses mechanical forces in endothelial, red blood cell and other cells in the cardiovascular system such as cardiac fibroblasts. Malfunction of Piezo1 can lead to a number of cardiovascular diseases. ARCHER allocation allowed us to use computational methods in order to study the spatial organization of Piezo1 in model membranes that resemble the endothelial membrane in which Piezo1 functions; this study wouldn’t have been possible without the resources we had from HECBioSim”

This work as also formed the basis of an FLF application for which Khalid is the external mentor.

This shows the utility of the consortium model in not only provided HPC time but also mentorship and advice which can be crucial for ECRs.